CPU · Atomli 0.1.4 · 8 threads
qc.rs PySCFLower is better
Molecules PBE / def2-SVP · RI-J · 30 AO sizes per CPU · logarithmic axes
Apple M4 Max Per SCF time 1.00 ms 10.0 ms 100 ms 1.00 s 10.0 s 100 s 10 20 50 100 200 500 1000 Atomic orbitals g2-H2 · 10 AO · qc.rs 11.4 ms; PySCF 20.4 ms. 1.8× faster. g2-H2 · 10 AO · qc.rs 11.4 ms; PySCF 20.4 ms. 1.8× faster. g2-CH2_s1A1d · 24 AO · qc.rs 6.77 ms; PySCF 17.0 ms. 2.5× faster. g2-CH2_s1A1d · 24 AO · qc.rs 6.77 ms; PySCF 17.0 ms. 2.5× faster. g2-NH3 · 29 AO · qc.rs 7.46 ms; PySCF 27.8 ms. 3.7× faster. g2-NH3 · 29 AO · qc.rs 7.46 ms; PySCF 27.8 ms. 3.7× faster. g2-CH4 · 34 AO · qc.rs 8.33 ms; PySCF 49.2 ms. 5.9× faster. g2-CH4 · 34 AO · qc.rs 8.33 ms; PySCF 49.2 ms. 5.9× faster. g2-HOCl · 37 AO · qc.rs 8.03 ms; PySCF 38.9 ms. 4.8× faster. g2-HOCl · 37 AO · qc.rs 8.03 ms; PySCF 38.9 ms. 4.8× faster. g2-C2H2 · 38 AO · qc.rs 7.61 ms; PySCF 51.5 ms. 6.8× faster. g2-C2H2 · 38 AO · qc.rs 7.61 ms; PySCF 51.5 ms. 6.8× faster. g2-ClNO · 46 AO · qc.rs 8.21 ms; PySCF 49.2 ms. 6× faster. g2-ClNO · 46 AO · qc.rs 8.21 ms; PySCF 49.2 ms. 6× faster. g2-C2H4 · 48 AO · qc.rs 12.1 ms; PySCF 90.1 ms. 7.5× faster. g2-C2H4 · 48 AO · qc.rs 12.1 ms; PySCF 90.1 ms. 7.5× faster. g2-H3CNH2 · 53 AO · qc.rs 14.4 ms; PySCF 113 ms. 7.8× faster. g2-H3CNH2 · 53 AO · qc.rs 14.4 ms; PySCF 113 ms. 7.8× faster. g2-CH3CN · 57 AO · qc.rs 13.5 ms; PySCF 83.7 ms. 6.2× faster. g2-CH3CN · 57 AO · qc.rs 13.5 ms; PySCF 83.7 ms. 6.2× faster. g2-C2H6 · 58 AO · qc.rs 19.0 ms; PySCF 137 ms. 7.2× faster. g2-C2H6 · 58 AO · qc.rs 19.0 ms; PySCF 137 ms. 7.2× faster. g2-H2CCHCl · 61 AO · qc.rs 18.0 ms; PySCF 97.4 ms. 5.4× faster. g2-H2CCHCl · 61 AO · qc.rs 18.0 ms; PySCF 97.4 ms. 5.4× faster. g2-CH2SCH2 · 66 AO · qc.rs 20.0 ms; PySCF 118 ms. 5.9× faster. g2-CH2SCH2 · 66 AO · qc.rs 20.0 ms; PySCF 118 ms. 5.9× faster. g2-CH3CH2Cl · 71 AO · qc.rs 21.9 ms; PySCF 161 ms. 7.3× faster. g2-CH3CH2Cl · 71 AO · qc.rs 21.9 ms; PySCF 161 ms. 7.3× faster. g2-HCCl3 · 73 AO · qc.rs 20.6 ms; PySCF 158 ms. 7.7× faster. g2-HCCl3 · 73 AO · qc.rs 20.6 ms; PySCF 158 ms. 7.7× faster. g2-CH3CH2SH · 76 AO · qc.rs 25.8 ms; PySCF 186 ms. 7.2× faster. g2-CH3CH2SH · 76 AO · qc.rs 25.8 ms; PySCF 186 ms. 7.2× faster. g2-C2H6NH · 77 AO · qc.rs 27.1 ms; PySCF 192 ms. 7.1× faster. g2-C2H6NH · 77 AO · qc.rs 27.1 ms; PySCF 192 ms. 7.1× faster. g2-CH3CONH2 · 81 AO · qc.rs 24.8 ms; PySCF 177 ms. 7.2× faster. g2-CH3CONH2 · 81 AO · qc.rs 24.8 ms; PySCF 177 ms. 7.2× faster. g2-2-butyne · 86 AO · qc.rs 29.6 ms; PySCF 274 ms. 9.3× faster. g2-2-butyne · 86 AO · qc.rs 29.6 ms; PySCF 274 ms. 9.3× faster. g2-C4H4S · 94 AO · qc.rs 34.3 ms; PySCF 249 ms. 7.3× faster. g2-C4H4S · 94 AO · qc.rs 34.3 ms; PySCF 249 ms. 7.3× faster. g2-C2Cl4 · 100 AO · qc.rs 35.3 ms; PySCF 279 ms. 7.9× faster. g2-C2Cl4 · 100 AO · qc.rs 35.3 ms; PySCF 279 ms. 7.9× faster. g2-isobutane · 106 AO · qc.rs 54.2 ms; PySCF 387 ms. 7.1× faster. g2-isobutane · 106 AO · qc.rs 54.2 ms; PySCF 387 ms. 7.1× faster. g2-C6H6 · 114 AO · qc.rs 52.2 ms; PySCF 380 ms. 7.3× faster. g2-C6H6 · 114 AO · qc.rs 52.2 ms; PySCF 380 ms. 7.3× faster. alkane-6 · 154 AO · qc.rs 102 ms; PySCF 776 ms. 7.6× faster. alkane-6 · 154 AO · qc.rs 102 ms; PySCF 776 ms. 7.6× faster. alkane-8 · 202 AO · qc.rs 166 ms; PySCF 1.41 s. 8.5× faster. alkane-8 · 202 AO · qc.rs 166 ms; PySCF 1.41 s. 8.5× faster. caffeine · 246 AO · qc.rs 236 ms; PySCF 2.34 s. 9.9× faster. caffeine · 246 AO · qc.rs 236 ms; PySCF 2.34 s. 9.9× faster. alkane-10 · 250 AO · qc.rs 253 ms; PySCF 2.42 s. 9.6× faster. alkane-10 · 250 AO · qc.rs 253 ms; PySCF 2.42 s. 9.6× faster. alkane-12 · 298 AO · qc.rs 397 ms; PySCF 4.25 s. 10.7× faster. alkane-12 · 298 AO · qc.rs 397 ms; PySCF 4.25 s. 10.7× faster. alkane-16 · 394 AO · qc.rs 646 ms; PySCF 10.3 s. 16× faster. alkane-16 · 394 AO · qc.rs 646 ms; PySCF 10.3 s. 16× faster. alkane-24 · 586 AO · qc.rs 1.25 s; PySCF 19.8 s. 15.8× faster. alkane-24 · 586 AO · qc.rs 1.25 s; PySCF 19.8 s. 15.8× faster. Peak host RSS 95.4 MiB 954 MiB 9.31 GiB 10 20 50 100 200 500 1000 Atomic orbitals g2-H2 · 10 AO · qc.rs 108 MiB; PySCF 125 MiB. 14% less memory. g2-H2 · 10 AO · qc.rs 108 MiB; PySCF 125 MiB. 14% less memory. g2-CH2_s1A1d · 24 AO · qc.rs 142 MiB; PySCF 156 MiB. 9% less memory. g2-CH2_s1A1d · 24 AO · qc.rs 142 MiB; PySCF 156 MiB. 9% less memory. g2-NH3 · 29 AO · qc.rs 160 MiB; PySCF 183 MiB. 12% less memory. g2-NH3 · 29 AO · qc.rs 160 MiB; PySCF 183 MiB. 12% less memory. g2-CH4 · 34 AO · qc.rs 176 MiB; PySCF 215 MiB. 18% less memory. g2-CH4 · 34 AO · qc.rs 176 MiB; PySCF 215 MiB. 18% less memory. g2-HOCl · 37 AO · qc.rs 175 MiB; PySCF 202 MiB. 14% less memory. g2-HOCl · 37 AO · qc.rs 175 MiB; PySCF 202 MiB. 14% less memory. g2-C2H2 · 38 AO · qc.rs 175 MiB; PySCF 207 MiB. 16% less memory. g2-C2H2 · 38 AO · qc.rs 175 MiB; PySCF 207 MiB. 16% less memory. g2-ClNO · 46 AO · qc.rs 170 MiB; PySCF 231 MiB. 26% less memory. g2-ClNO · 46 AO · qc.rs 170 MiB; PySCF 231 MiB. 26% less memory. g2-C2H4 · 48 AO · qc.rs 198 MiB; PySCF 279 MiB. 29% less memory. g2-C2H4 · 48 AO · qc.rs 198 MiB; PySCF 279 MiB. 29% less memory. g2-H3CNH2 · 53 AO · qc.rs 207 MiB; PySCF 300 MiB. 31% less memory. g2-H3CNH2 · 53 AO · qc.rs 207 MiB; PySCF 300 MiB. 31% less memory. g2-CH3CN · 57 AO · qc.rs 213 MiB; PySCF 312 MiB. 32% less memory. g2-CH3CN · 57 AO · qc.rs 213 MiB; PySCF 312 MiB. 32% less memory. g2-C2H6 · 58 AO · qc.rs 225 MiB; PySCF 335 MiB. 33% less memory. g2-C2H6 · 58 AO · qc.rs 225 MiB; PySCF 335 MiB. 33% less memory. g2-H2CCHCl · 61 AO · qc.rs 226 MiB; PySCF 329 MiB. 31% less memory. g2-H2CCHCl · 61 AO · qc.rs 226 MiB; PySCF 329 MiB. 31% less memory. g2-CH2SCH2 · 66 AO · qc.rs 215 MiB; PySCF 359 MiB. 40% less memory. g2-CH2SCH2 · 66 AO · qc.rs 215 MiB; PySCF 359 MiB. 40% less memory. g2-CH3CH2Cl · 71 AO · qc.rs 237 MiB; PySCF 378 MiB. 37% less memory. g2-CH3CH2Cl · 71 AO · qc.rs 237 MiB; PySCF 378 MiB. 37% less memory. g2-HCCl3 · 73 AO · qc.rs 202 MiB; PySCF 388 MiB. 48% less memory. g2-HCCl3 · 73 AO · qc.rs 202 MiB; PySCF 388 MiB. 48% less memory. g2-CH3CH2SH · 76 AO · qc.rs 222 MiB; PySCF 396 MiB. 44% less memory. g2-CH3CH2SH · 76 AO · qc.rs 222 MiB; PySCF 396 MiB. 44% less memory. g2-C2H6NH · 77 AO · qc.rs 243 MiB; PySCF 402 MiB. 39% less memory. g2-C2H6NH · 77 AO · qc.rs 243 MiB; PySCF 402 MiB. 39% less memory. g2-CH3CONH2 · 81 AO · qc.rs 253 MiB; PySCF 405 MiB. 37% less memory. g2-CH3CONH2 · 81 AO · qc.rs 253 MiB; PySCF 405 MiB. 37% less memory. g2-2-butyne · 86 AO · qc.rs 213 MiB; PySCF 426 MiB. 50% less memory. g2-2-butyne · 86 AO · qc.rs 213 MiB; PySCF 426 MiB. 50% less memory. g2-C4H4S · 94 AO · qc.rs 257 MiB; PySCF 467 MiB. 45% less memory. g2-C4H4S · 94 AO · qc.rs 257 MiB; PySCF 467 MiB. 45% less memory. g2-C2Cl4 · 100 AO · qc.rs 250 MiB; PySCF 517 MiB. 52% less memory. g2-C2Cl4 · 100 AO · qc.rs 250 MiB; PySCF 517 MiB. 52% less memory. g2-isobutane · 106 AO · qc.rs 269 MiB; PySCF 497 MiB. 46% less memory. g2-isobutane · 106 AO · qc.rs 269 MiB; PySCF 497 MiB. 46% less memory. g2-C6H6 · 114 AO · qc.rs 265 MiB; PySCF 512 MiB. 48% less memory. g2-C6H6 · 114 AO · qc.rs 265 MiB; PySCF 512 MiB. 48% less memory. alkane-6 · 154 AO · qc.rs 277 MiB; PySCF 673 MiB. 59% less memory. alkane-6 · 154 AO · qc.rs 277 MiB; PySCF 673 MiB. 59% less memory. alkane-8 · 202 AO · qc.rs 353 MiB; PySCF 838 MiB. 58% less memory. alkane-8 · 202 AO · qc.rs 353 MiB; PySCF 838 MiB. 58% less memory. caffeine · 246 AO · qc.rs 386 MiB; PySCF 1.02e+3 MiB. 62% less memory. caffeine · 246 AO · qc.rs 386 MiB; PySCF 1.02e+3 MiB. 62% less memory. alkane-10 · 250 AO · qc.rs 387 MiB; PySCF 1.02e+3 MiB. 62% less memory. alkane-10 · 250 AO · qc.rs 387 MiB; PySCF 1.02e+3 MiB. 62% less memory. alkane-12 · 298 AO · qc.rs 406 MiB; PySCF 1.33 GiB. 70% less memory. alkane-12 · 298 AO · qc.rs 406 MiB; PySCF 1.33 GiB. 70% less memory. alkane-16 · 394 AO · qc.rs 501 MiB; PySCF 1.54 GiB. 68% less memory. alkane-16 · 394 AO · qc.rs 501 MiB; PySCF 1.54 GiB. 68% less memory. alkane-24 · 586 AO · qc.rs 738 MiB; PySCF 2.83 GiB. 75% less memory. alkane-24 · 586 AO · qc.rs 738 MiB; PySCF 2.83 GiB. 75% less memory.
Apple M3 Ultra Per SCF time 1.00 ms 10.0 ms 100 ms 1.00 s 10.0 s 100 s 10 20 50 100 200 500 1000 Atomic orbitals g2-H2 · 10 AO · qc.rs 13.7 ms; PySCF 22.1 ms. 1.6× faster. g2-H2 · 10 AO · qc.rs 13.7 ms; PySCF 22.1 ms. 1.6× faster. g2-CH2_s1A1d · 24 AO · qc.rs 8.21 ms; PySCF 20.5 ms. 2.5× faster. g2-CH2_s1A1d · 24 AO · qc.rs 8.21 ms; PySCF 20.5 ms. 2.5× faster. g2-NH3 · 29 AO · qc.rs 9.41 ms; PySCF 35.5 ms. 3.8× faster. g2-NH3 · 29 AO · qc.rs 9.41 ms; PySCF 35.5 ms. 3.8× faster. g2-CH4 · 34 AO · qc.rs 10.8 ms; PySCF 60.3 ms. 5.6× faster. g2-CH4 · 34 AO · qc.rs 10.8 ms; PySCF 60.3 ms. 5.6× faster. g2-HOCl · 37 AO · qc.rs 10.3 ms; PySCF 47.6 ms. 4.6× faster. g2-HOCl · 37 AO · qc.rs 10.3 ms; PySCF 47.6 ms. 4.6× faster. g2-C2H2 · 38 AO · qc.rs 10.8 ms; PySCF 61.2 ms. 5.6× faster. g2-C2H2 · 38 AO · qc.rs 10.8 ms; PySCF 61.2 ms. 5.6× faster. g2-ClNO · 46 AO · qc.rs 10.8 ms; PySCF 54.1 ms. 5× faster. g2-ClNO · 46 AO · qc.rs 10.8 ms; PySCF 54.1 ms. 5× faster. g2-C2H4 · 48 AO · qc.rs 16.7 ms; PySCF 98.4 ms. 5.9× faster. g2-C2H4 · 48 AO · qc.rs 16.7 ms; PySCF 98.4 ms. 5.9× faster. g2-H3CNH2 · 53 AO · qc.rs 19.3 ms; PySCF 122 ms. 6.3× faster. g2-H3CNH2 · 53 AO · qc.rs 19.3 ms; PySCF 122 ms. 6.3× faster. g2-CH3CN · 57 AO · qc.rs 18.1 ms; PySCF 91.3 ms. 5× faster. g2-CH3CN · 57 AO · qc.rs 18.1 ms; PySCF 91.3 ms. 5× faster. g2-C2H6 · 58 AO · qc.rs 25.2 ms; PySCF 148 ms. 5.9× faster. g2-C2H6 · 58 AO · qc.rs 25.2 ms; PySCF 148 ms. 5.9× faster. g2-H2CCHCl · 61 AO · qc.rs 23.9 ms; PySCF 105 ms. 4.4× faster. g2-H2CCHCl · 61 AO · qc.rs 23.9 ms; PySCF 105 ms. 4.4× faster. g2-CH2SCH2 · 66 AO · qc.rs 26.6 ms; PySCF 127 ms. 4.8× faster. g2-CH2SCH2 · 66 AO · qc.rs 26.6 ms; PySCF 127 ms. 4.8× faster. g2-CH3CH2Cl · 71 AO · qc.rs 28.8 ms; PySCF 170 ms. 5.9× faster. g2-CH3CH2Cl · 71 AO · qc.rs 28.8 ms; PySCF 170 ms. 5.9× faster. g2-HCCl3 · 73 AO · qc.rs 27.7 ms; PySCF 168 ms. 6.1× faster. g2-HCCl3 · 73 AO · qc.rs 27.7 ms; PySCF 168 ms. 6.1× faster. g2-CH3CH2SH · 76 AO · qc.rs 32.9 ms; PySCF 198 ms. 6× faster. g2-CH3CH2SH · 76 AO · qc.rs 32.9 ms; PySCF 198 ms. 6× faster. g2-CH3CONH2 · 81 AO · qc.rs 31.9 ms; PySCF 190 ms. 5.9× faster. g2-CH3CONH2 · 81 AO · qc.rs 31.9 ms; PySCF 190 ms. 5.9× faster. g2-2-butyne · 86 AO · qc.rs 38.3 ms; PySCF 314 ms. 8.2× faster. g2-2-butyne · 86 AO · qc.rs 38.3 ms; PySCF 314 ms. 8.2× faster. g2-C4H4S · 94 AO · qc.rs 43.7 ms; PySCF 282 ms. 6.4× faster. g2-C4H4S · 94 AO · qc.rs 43.7 ms; PySCF 282 ms. 6.4× faster. g2-C2Cl4 · 100 AO · qc.rs 44.7 ms; PySCF 306 ms. 6.8× faster. g2-C2Cl4 · 100 AO · qc.rs 44.7 ms; PySCF 306 ms. 6.8× faster. g2-isobutane · 106 AO · qc.rs 68.6 ms; PySCF 429 ms. 6.3× faster. g2-isobutane · 106 AO · qc.rs 68.6 ms; PySCF 429 ms. 6.3× faster. g2-C6H6 · 114 AO · qc.rs 66.1 ms; PySCF 425 ms. 6.4× faster. g2-C6H6 · 114 AO · qc.rs 66.1 ms; PySCF 425 ms. 6.4× faster. alkane-6 · 154 AO · qc.rs 128 ms; PySCF 835 ms. 6.5× faster. alkane-6 · 154 AO · qc.rs 128 ms; PySCF 835 ms. 6.5× faster. alkane-8 · 202 AO · qc.rs 202 ms; PySCF 1.52 s. 7.5× faster. alkane-8 · 202 AO · qc.rs 202 ms; PySCF 1.52 s. 7.5× faster. caffeine · 246 AO · qc.rs 281 ms; PySCF 2.51 s. 8.9× faster. caffeine · 246 AO · qc.rs 281 ms; PySCF 2.51 s. 8.9× faster. alkane-10 · 250 AO · qc.rs 304 ms; PySCF 2.59 s. 8.5× faster. alkane-10 · 250 AO · qc.rs 304 ms; PySCF 2.59 s. 8.5× faster. alkane-12 · 298 AO · qc.rs 471 ms; PySCF 4.62 s. 9.8× faster. alkane-12 · 298 AO · qc.rs 471 ms; PySCF 4.62 s. 9.8× faster. alkane-16 · 394 AO · qc.rs 760 ms; PySCF 11.6 s. 15.2× faster. alkane-16 · 394 AO · qc.rs 760 ms; PySCF 11.6 s. 15.2× faster. alkane-20 · 490 AO · qc.rs 1.10 s; PySCF 16.5 s. 15× faster. alkane-20 · 490 AO · qc.rs 1.10 s; PySCF 16.5 s. 15× faster. alkane-24 · 586 AO · qc.rs 1.48 s; PySCF 22.0 s. 14.9× faster. alkane-24 · 586 AO · qc.rs 1.48 s; PySCF 22.0 s. 14.9× faster. Peak host RSS 95.4 MiB 954 MiB 9.31 GiB 10 20 50 100 200 500 1000 Atomic orbitals g2-H2 · 10 AO · qc.rs 139 MiB; PySCF 123 MiB. g2-H2 · 10 AO · qc.rs 139 MiB; PySCF 123 MiB. g2-CH2_s1A1d · 24 AO · qc.rs 208 MiB; PySCF 149 MiB. g2-CH2_s1A1d · 24 AO · qc.rs 208 MiB; PySCF 149 MiB. g2-NH3 · 29 AO · qc.rs 218 MiB; PySCF 225 MiB. 3% less memory. g2-NH3 · 29 AO · qc.rs 218 MiB; PySCF 225 MiB. 3% less memory. g2-CH4 · 34 AO · qc.rs 241 MiB; PySCF 243 MiB. <1% less memory. g2-CH4 · 34 AO · qc.rs 241 MiB; PySCF 243 MiB. <1% less memory. g2-HOCl · 37 AO · qc.rs 224 MiB; PySCF 259 MiB. 13% less memory. g2-HOCl · 37 AO · qc.rs 224 MiB; PySCF 259 MiB. 13% less memory. g2-C2H2 · 38 AO · qc.rs 271 MiB; PySCF 247 MiB. g2-C2H2 · 38 AO · qc.rs 271 MiB; PySCF 247 MiB. g2-ClNO · 46 AO · qc.rs 281 MiB; PySCF 335 MiB. 16% less memory. g2-ClNO · 46 AO · qc.rs 281 MiB; PySCF 335 MiB. 16% less memory. g2-C2H4 · 48 AO · qc.rs 354 MiB; PySCF 363 MiB. 3% less memory. g2-C2H4 · 48 AO · qc.rs 354 MiB; PySCF 363 MiB. 3% less memory. g2-H3CNH2 · 53 AO · qc.rs 336 MiB; PySCF 394 MiB. 15% less memory. g2-H3CNH2 · 53 AO · qc.rs 336 MiB; PySCF 394 MiB. 15% less memory. g2-CH3CN · 57 AO · qc.rs 368 MiB; PySCF 434 MiB. 15% less memory. g2-CH3CN · 57 AO · qc.rs 368 MiB; PySCF 434 MiB. 15% less memory. g2-C2H6 · 58 AO · qc.rs 367 MiB; PySCF 404 MiB. 9% less memory. g2-C2H6 · 58 AO · qc.rs 367 MiB; PySCF 404 MiB. 9% less memory. g2-H2CCHCl · 61 AO · qc.rs 347 MiB; PySCF 401 MiB. 13% less memory. g2-H2CCHCl · 61 AO · qc.rs 347 MiB; PySCF 401 MiB. 13% less memory. g2-CH2SCH2 · 66 AO · qc.rs 464 MiB; PySCF 435 MiB. g2-CH2SCH2 · 66 AO · qc.rs 464 MiB; PySCF 435 MiB. g2-CH3CH2Cl · 71 AO · qc.rs 412 MiB; PySCF 483 MiB. 15% less memory. g2-CH3CH2Cl · 71 AO · qc.rs 412 MiB; PySCF 483 MiB. 15% less memory. g2-HCCl3 · 73 AO · qc.rs 387 MiB; PySCF 473 MiB. 18% less memory. g2-HCCl3 · 73 AO · qc.rs 387 MiB; PySCF 473 MiB. 18% less memory. g2-CH3CH2SH · 76 AO · qc.rs 422 MiB; PySCF 523 MiB. 19% less memory. g2-CH3CH2SH · 76 AO · qc.rs 422 MiB; PySCF 523 MiB. 19% less memory. g2-CH3CONH2 · 81 AO · qc.rs 463 MiB; PySCF 561 MiB. 17% less memory. g2-CH3CONH2 · 81 AO · qc.rs 463 MiB; PySCF 561 MiB. 17% less memory. g2-2-butyne · 86 AO · qc.rs 452 MiB; PySCF 602 MiB. 25% less memory. g2-2-butyne · 86 AO · qc.rs 452 MiB; PySCF 602 MiB. 25% less memory. g2-C4H4S · 94 AO · qc.rs 529 MiB; PySCF 638 MiB. 17% less memory. g2-C4H4S · 94 AO · qc.rs 529 MiB; PySCF 638 MiB. 17% less memory. g2-C2Cl4 · 100 AO · qc.rs 435 MiB; PySCF 607 MiB. 28% less memory. g2-C2Cl4 · 100 AO · qc.rs 435 MiB; PySCF 607 MiB. 28% less memory. g2-isobutane · 106 AO · qc.rs 519 MiB; PySCF 618 MiB. 16% less memory. g2-isobutane · 106 AO · qc.rs 519 MiB; PySCF 618 MiB. 16% less memory. g2-C6H6 · 114 AO · qc.rs 736 MiB; PySCF 617 MiB. g2-C6H6 · 114 AO · qc.rs 736 MiB; PySCF 617 MiB. alkane-6 · 154 AO · qc.rs 822 MiB; PySCF 936 MiB. 12% less memory. alkane-6 · 154 AO · qc.rs 822 MiB; PySCF 936 MiB. 12% less memory. alkane-8 · 202 AO · qc.rs 903 MiB; PySCF 1.14 GiB. 23% less memory. alkane-8 · 202 AO · qc.rs 903 MiB; PySCF 1.14 GiB. 23% less memory. caffeine · 246 AO · qc.rs 1.05 GiB; PySCF 1.39 GiB. 24% less memory. caffeine · 246 AO · qc.rs 1.05 GiB; PySCF 1.39 GiB. 24% less memory. alkane-10 · 250 AO · qc.rs 1.08 GiB; PySCF 1.38 GiB. 22% less memory. alkane-10 · 250 AO · qc.rs 1.08 GiB; PySCF 1.38 GiB. 22% less memory. alkane-12 · 298 AO · qc.rs 1.07 GiB; PySCF 1.72 GiB. 38% less memory. alkane-12 · 298 AO · qc.rs 1.07 GiB; PySCF 1.72 GiB. 38% less memory. alkane-16 · 394 AO · qc.rs 1.21 GiB; PySCF 2.05 GiB. 41% less memory. alkane-16 · 394 AO · qc.rs 1.21 GiB; PySCF 2.05 GiB. 41% less memory. alkane-20 · 490 AO · qc.rs 1.39 GiB; PySCF 2.53 GiB. 45% less memory. alkane-20 · 490 AO · qc.rs 1.39 GiB; PySCF 2.53 GiB. 45% less memory. alkane-24 · 586 AO · qc.rs 1.56 GiB; PySCF 2.90 GiB. 46% less memory. alkane-24 · 586 AO · qc.rs 1.56 GiB; PySCF 2.90 GiB. 46% less memory.
AMD EPYC 7702P Per SCF time 1.00 ms 10.0 ms 100 ms 1.00 s 10.0 s 100 s 10 20 50 100 200 500 1000 Atomic orbitals g2-H2 · 10 AO · qc.rs 114 ms; PySCF 39.1 ms. 2.9× slower. g2-H2 · 10 AO · qc.rs 114 ms; PySCF 39.1 ms. 2.9× slower. g2-CH2_s1A1d · 24 AO · qc.rs 65.0 ms; PySCF 29.4 ms. 2.2× slower. g2-CH2_s1A1d · 24 AO · qc.rs 65.0 ms; PySCF 29.4 ms. 2.2× slower. g2-NH3 · 29 AO · qc.rs 63.2 ms; PySCF 51.5 ms. 1.2× slower. g2-NH3 · 29 AO · qc.rs 63.2 ms; PySCF 51.5 ms. 1.2× slower. g2-CH4 · 34 AO · qc.rs 42.2 ms; PySCF 91.2 ms. 2.2× faster. g2-CH4 · 34 AO · qc.rs 42.2 ms; PySCF 91.2 ms. 2.2× faster. g2-HOCl · 37 AO · qc.rs 42.1 ms; PySCF 61.2 ms. 1.5× faster. g2-HOCl · 37 AO · qc.rs 42.1 ms; PySCF 61.2 ms. 1.5× faster. g2-C2H2 · 38 AO · qc.rs 37.5 ms; PySCF 84.0 ms. 2.2× faster. g2-C2H2 · 38 AO · qc.rs 37.5 ms; PySCF 84.0 ms. 2.2× faster. g2-ClNO · 46 AO · qc.rs 35.3 ms; PySCF 65.3 ms. 1.8× faster. g2-ClNO · 46 AO · qc.rs 35.3 ms; PySCF 65.3 ms. 1.8× faster. g2-C2H4 · 48 AO · qc.rs 71.8 ms; PySCF 128 ms. 1.8× faster. g2-C2H4 · 48 AO · qc.rs 71.8 ms; PySCF 128 ms. 1.8× faster. g2-H3CNH2 · 53 AO · qc.rs 65.0 ms; PySCF 153 ms. 2.4× faster. g2-H3CNH2 · 53 AO · qc.rs 65.0 ms; PySCF 153 ms. 2.4× faster. g2-CH3CN · 57 AO · qc.rs 61.4 ms; PySCF 104 ms. 1.7× faster. g2-CH3CN · 57 AO · qc.rs 61.4 ms; PySCF 104 ms. 1.7× faster. g2-H2CCHCl · 61 AO · qc.rs 86.7 ms; PySCF 117 ms. 1.4× faster. g2-H2CCHCl · 61 AO · qc.rs 86.7 ms; PySCF 117 ms. 1.4× faster. g2-CH2SCH2 · 66 AO · qc.rs 96.7 ms; PySCF 148 ms. 1.5× faster. g2-CH2SCH2 · 66 AO · qc.rs 96.7 ms; PySCF 148 ms. 1.5× faster. g2-CH3CH2Cl · 71 AO · qc.rs 118 ms; PySCF 219 ms. 1.9× faster. g2-CH3CH2Cl · 71 AO · qc.rs 118 ms; PySCF 219 ms. 1.9× faster. g2-HCCl3 · 73 AO · qc.rs 94.8 ms; PySCF 183 ms. 1.9× faster. g2-HCCl3 · 73 AO · qc.rs 94.8 ms; PySCF 183 ms. 1.9× faster. g2-CH3CH2SH · 76 AO · qc.rs 121 ms; PySCF 212 ms. 1.8× faster. g2-CH3CH2SH · 76 AO · qc.rs 121 ms; PySCF 212 ms. 1.8× faster. g2-CH3CONH2 · 81 AO · qc.rs 132 ms; PySCF 184 ms. 1.4× faster. g2-CH3CONH2 · 81 AO · qc.rs 132 ms; PySCF 184 ms. 1.4× faster. g2-2-butyne · 86 AO · qc.rs 149 ms; PySCF 299 ms. 2× faster. g2-2-butyne · 86 AO · qc.rs 149 ms; PySCF 299 ms. 2× faster. g2-C4H4S · 94 AO · qc.rs 156 ms; PySCF 293 ms. 1.9× faster. g2-C4H4S · 94 AO · qc.rs 156 ms; PySCF 293 ms. 1.9× faster. g2-C2Cl4 · 100 AO · qc.rs 140 ms; PySCF 270 ms. 1.9× faster. g2-C2Cl4 · 100 AO · qc.rs 140 ms; PySCF 270 ms. 1.9× faster. g2-isobutane · 106 AO · qc.rs 245 ms; PySCF 352 ms. 1.4× faster. g2-isobutane · 106 AO · qc.rs 245 ms; PySCF 352 ms. 1.4× faster. g2-C6H6 · 114 AO · qc.rs 224 ms; PySCF 347 ms. 1.5× faster. g2-C6H6 · 114 AO · qc.rs 224 ms; PySCF 347 ms. 1.5× faster. alkane-6 · 154 AO · qc.rs 430 ms; PySCF 622 ms. 1.4× faster. alkane-6 · 154 AO · qc.rs 430 ms; PySCF 622 ms. 1.4× faster. alkane-8 · 202 AO · qc.rs 680 ms; PySCF 1.10 s. 1.6× faster. alkane-8 · 202 AO · qc.rs 680 ms; PySCF 1.10 s. 1.6× faster. caffeine · 246 AO · qc.rs 922 ms; PySCF 1.68 s. 1.8× faster. caffeine · 246 AO · qc.rs 922 ms; PySCF 1.68 s. 1.8× faster. alkane-10 · 250 AO · qc.rs 1.04 s; PySCF 2.03 s. 1.9× faster. alkane-10 · 250 AO · qc.rs 1.04 s; PySCF 2.03 s. 1.9× faster. alkane-12 · 298 AO · qc.rs 1.43 s; PySCF 3.13 s. 2.2× faster. alkane-12 · 298 AO · qc.rs 1.43 s; PySCF 3.13 s. 2.2× faster. alkane-16 · 394 AO · qc.rs 2.30 s; PySCF 6.37 s. 2.8× faster. alkane-16 · 394 AO · qc.rs 2.30 s; PySCF 6.37 s. 2.8× faster. alkane-20 · 490 AO · qc.rs 3.18 s; PySCF 8.50 s. 2.7× faster. alkane-20 · 490 AO · qc.rs 3.18 s; PySCF 8.50 s. 2.7× faster. alkane-24 · 586 AO · qc.rs 4.28 s; PySCF 11.5 s. 2.7× faster. alkane-24 · 586 AO · qc.rs 4.28 s; PySCF 11.5 s. 2.7× faster. alkane-40 · 970 AO · qc.rs 10.3 s; PySCF 23.3 s. 2.3× faster. alkane-40 · 970 AO · qc.rs 10.3 s; PySCF 23.3 s. 2.3× faster. Peak host RSS 95.4 MiB 954 MiB 9.31 GiB 10 20 50 100 200 500 1000 Atomic orbitals g2-H2 · 10 AO · qc.rs 143 MiB; PySCF 136 MiB. g2-H2 · 10 AO · qc.rs 143 MiB; PySCF 136 MiB. g2-CH2_s1A1d · 24 AO · qc.rs 185 MiB; PySCF 165 MiB. g2-CH2_s1A1d · 24 AO · qc.rs 185 MiB; PySCF 165 MiB. g2-NH3 · 29 AO · qc.rs 199 MiB; PySCF 185 MiB. g2-NH3 · 29 AO · qc.rs 199 MiB; PySCF 185 MiB. g2-CH4 · 34 AO · qc.rs 199 MiB; PySCF 208 MiB. 4% less memory. g2-CH4 · 34 AO · qc.rs 199 MiB; PySCF 208 MiB. 4% less memory. g2-HOCl · 37 AO · qc.rs 204 MiB; PySCF 197 MiB. g2-HOCl · 37 AO · qc.rs 204 MiB; PySCF 197 MiB. g2-C2H2 · 38 AO · qc.rs 197 MiB; PySCF 207 MiB. 5% less memory. g2-C2H2 · 38 AO · qc.rs 197 MiB; PySCF 207 MiB. 5% less memory. g2-ClNO · 46 AO · qc.rs 198 MiB; PySCF 223 MiB. 11% less memory. g2-ClNO · 46 AO · qc.rs 198 MiB; PySCF 223 MiB. 11% less memory. g2-C2H4 · 48 AO · qc.rs 208 MiB; PySCF 267 MiB. 22% less memory. g2-C2H4 · 48 AO · qc.rs 208 MiB; PySCF 267 MiB. 22% less memory. g2-H3CNH2 · 53 AO · qc.rs 216 MiB; PySCF 281 MiB. 23% less memory. g2-H3CNH2 · 53 AO · qc.rs 216 MiB; PySCF 281 MiB. 23% less memory. g2-CH3CN · 57 AO · qc.rs 212 MiB; PySCF 291 MiB. 27% less memory. g2-CH3CN · 57 AO · qc.rs 212 MiB; PySCF 291 MiB. 27% less memory. g2-H2CCHCl · 61 AO · qc.rs 226 MiB; PySCF 305 MiB. 26% less memory. g2-H2CCHCl · 61 AO · qc.rs 226 MiB; PySCF 305 MiB. 26% less memory. g2-CH2SCH2 · 66 AO · qc.rs 234 MiB; PySCF 323 MiB. 27% less memory. g2-CH2SCH2 · 66 AO · qc.rs 234 MiB; PySCF 323 MiB. 27% less memory. g2-CH3CH2Cl · 71 AO · qc.rs 248 MiB; PySCF 343 MiB. 28% less memory. g2-CH3CH2Cl · 71 AO · qc.rs 248 MiB; PySCF 343 MiB. 28% less memory. g2-HCCl3 · 73 AO · qc.rs 221 MiB; PySCF 345 MiB. 36% less memory. g2-HCCl3 · 73 AO · qc.rs 221 MiB; PySCF 345 MiB. 36% less memory. g2-CH3CH2SH · 76 AO · qc.rs 231 MiB; PySCF 361 MiB. 36% less memory. g2-CH3CH2SH · 76 AO · qc.rs 231 MiB; PySCF 361 MiB. 36% less memory. g2-CH3CONH2 · 81 AO · qc.rs 240 MiB; PySCF 379 MiB. 37% less memory. g2-CH3CONH2 · 81 AO · qc.rs 240 MiB; PySCF 379 MiB. 37% less memory. g2-2-butyne · 86 AO · qc.rs 230 MiB; PySCF 390 MiB. 41% less memory. g2-2-butyne · 86 AO · qc.rs 230 MiB; PySCF 390 MiB. 41% less memory. g2-C4H4S · 94 AO · qc.rs 258 MiB; PySCF 420 MiB. 39% less memory. g2-C4H4S · 94 AO · qc.rs 258 MiB; PySCF 420 MiB. 39% less memory. g2-C2Cl4 · 100 AO · qc.rs 254 MiB; PySCF 446 MiB. 43% less memory. g2-C2Cl4 · 100 AO · qc.rs 254 MiB; PySCF 446 MiB. 43% less memory. g2-isobutane · 106 AO · qc.rs 253 MiB; PySCF 465 MiB. 46% less memory. g2-isobutane · 106 AO · qc.rs 253 MiB; PySCF 465 MiB. 46% less memory. g2-C6H6 · 114 AO · qc.rs 266 MiB; PySCF 490 MiB. 46% less memory. g2-C6H6 · 114 AO · qc.rs 266 MiB; PySCF 490 MiB. 46% less memory. alkane-6 · 154 AO · qc.rs 245 MiB; PySCF 615 MiB. 60% less memory. alkane-6 · 154 AO · qc.rs 245 MiB; PySCF 615 MiB. 60% less memory. alkane-8 · 202 AO · qc.rs 304 MiB; PySCF 776 MiB. 61% less memory. alkane-8 · 202 AO · qc.rs 304 MiB; PySCF 776 MiB. 61% less memory. caffeine · 246 AO · qc.rs 307 MiB; PySCF 927 MiB. 67% less memory. caffeine · 246 AO · qc.rs 307 MiB; PySCF 927 MiB. 67% less memory. alkane-10 · 250 AO · qc.rs 360 MiB; PySCF 936 MiB. 62% less memory. alkane-10 · 250 AO · qc.rs 360 MiB; PySCF 936 MiB. 62% less memory. alkane-12 · 298 AO · qc.rs 371 MiB; PySCF 1.13 GiB. 68% less memory. alkane-12 · 298 AO · qc.rs 371 MiB; PySCF 1.13 GiB. 68% less memory. alkane-16 · 394 AO · qc.rs 526 MiB; PySCF 1.46 GiB. 65% less memory. alkane-16 · 394 AO · qc.rs 526 MiB; PySCF 1.46 GiB. 65% less memory. alkane-20 · 490 AO · qc.rs 715 MiB; PySCF 1.73 GiB. 60% less memory. alkane-20 · 490 AO · qc.rs 715 MiB; PySCF 1.73 GiB. 60% less memory. alkane-24 · 586 AO · qc.rs 722 MiB; PySCF 2.05 GiB. 66% less memory. alkane-24 · 586 AO · qc.rs 722 MiB; PySCF 2.05 GiB. 66% less memory. alkane-40 · 970 AO · qc.rs 1.67 GiB; PySCF 3.29 GiB. 49% less memory. alkane-40 · 970 AO · qc.rs 1.67 GiB; PySCF 3.29 GiB. 49% less memory.
Periodic cells PBE / GTH-DZVP-MOLOPT-SR · Γ sampling · 30 AO sizes per CPU · logarithmic axes
Apple M4 Max Per SCF time 10.0 ms 100 ms 1.00 s 10.0 s 100 s 1.00e+3 s 10 20 50 100 200 AO per cell lih-1 · 11 AO / cell · qc.rs 295 ms; PySCF 10.3 s. 34.9× faster. Grid 27 × 27 × 27; 17.0 ų. lih-1 · 11 AO / cell · qc.rs 295 ms; PySCF 10.3 s. 34.9× faster. Grid 27 × 27 × 27; 17.0 ų. lif-1 · 19 AO / cell · qc.rs 172 ms; PySCF 10.7 s. 62.1× faster. Grid 27 × 27 × 27; 16.2 ų. lif-1 · 19 AO / cell · qc.rs 172 ms; PySCF 10.7 s. 62.1× faster. Grid 27 × 27 × 27; 16.2 ų. lih-2 · 22 AO / cell · qc.rs 199 ms; PySCF 14.2 s. 71.6× faster. Grid 27 × 37 × 27; 34.0 ų. lih-2 · 22 AO / cell · qc.rs 199 ms; PySCF 14.2 s. 71.6× faster. Grid 27 × 37 × 27; 34.0 ų. oli2-1 · 25 AO / cell · qc.rs 230 ms; PySCF 15.6 s. 68× faster. Grid 29 × 29 × 29; 24.7 ų. oli2-1 · 25 AO / cell · qc.rs 230 ms; PySCF 15.6 s. 68× faster. Grid 29 × 29 × 29; 24.7 ų. c-1 · 26 AO / cell · qc.rs 52.4 ms; PySCF 2.27 s. 43.3× faster. Grid 23 × 23 × 23; 11.3 ų. c-1 · 26 AO / cell · qc.rs 52.4 ms; PySCF 2.27 s. 43.3× faster. Grid 23 × 23 × 23; 11.3 ų. kf-1 · 27 AO / cell · qc.rs 278 ms; PySCF 15.7 s. 56.5× faster. Grid 35 × 35 × 35; 38.5 ų. kf-1 · 27 AO / cell · qc.rs 278 ms; PySCF 15.7 s. 56.5× faster. Grid 35 × 35 × 35; 38.5 ų. lih-3 · 33 AO / cell · qc.rs 419 ms; PySCF 25.8 s. 61.7× faster. Grid 27 × 37 × 45; 51.1 ų. lih-3 · 33 AO / cell · qc.rs 419 ms; PySCF 25.8 s. 61.7× faster. Grid 27 × 37 × 45; 51.1 ų. lif-2 · 38 AO / cell · qc.rs 212 ms; PySCF 14.6 s. 68.8× faster. Grid 27 × 37 × 27; 32.5 ų. lif-2 · 38 AO / cell · qc.rs 212 ms; PySCF 14.6 s. 68.8× faster. Grid 27 × 37 × 27; 32.5 ų. lih-4 · 44 AO / cell · qc.rs 412 ms; PySCF 26.9 s. 65.4× faster. Grid 37 × 37 × 37; 68.1 ų. lih-4 · 44 AO / cell · qc.rs 412 ms; PySCF 26.9 s. 65.4× faster. Grid 37 × 37 × 37; 68.1 ų. c-2 · 52 AO / cell · qc.rs 67.4 ms; PySCF 2.40 s. 35.6× faster. Grid 23 × 33 × 23; 22.7 ų. c-2 · 52 AO / cell · qc.rs 67.4 ms; PySCF 2.40 s. 35.6× faster. Grid 23 × 33 × 23; 22.7 ų. lih-5 · 55 AO / cell · qc.rs 697 ms; PySCF 45.5 s. 65.3× faster. Grid 37 × 45 × 45; 85.1 ų. lih-5 · 55 AO / cell · qc.rs 697 ms; PySCF 45.5 s. 65.3× faster. Grid 37 × 45 × 45; 85.1 ų. lif-3 · 57 AO / cell · qc.rs 435 ms; PySCF 25.9 s. 59.4× faster. Grid 27 × 37 × 43; 48.7 ų. lif-3 · 57 AO / cell · qc.rs 435 ms; PySCF 25.9 s. 59.4× faster. Grid 27 × 37 × 43; 48.7 ų. lih-6 · 66 AO / cell · qc.rs 943 ms; PySCF 57.3 s. 60.7× faster. Grid 45 × 45 × 45; 102.1 ų. lih-6 · 66 AO / cell · qc.rs 943 ms; PySCF 57.3 s. 60.7× faster. Grid 45 × 45 × 45; 102.1 ų. lif-4 · 76 AO / cell · qc.rs 393 ms; PySCF 28.1 s. 71.4× faster. Grid 37 × 37 × 37; 65.0 ų. lif-4 · 76 AO / cell · qc.rs 393 ms; PySCF 28.1 s. 71.4× faster. Grid 37 × 37 × 37; 65.0 ų. lih-7 · 77 AO / cell · qc.rs 952 ms; PySCF 57.9 s. 60.8× faster. Grid 45 × 45 × 45; 119.1 ų. lih-7 · 77 AO / cell · qc.rs 952 ms; PySCF 57.9 s. 60.8× faster. Grid 45 × 45 × 45; 119.1 ų. c-3 · 78 AO / cell · qc.rs 155 ms; PySCF 5.16 s. 33.3× faster. Grid 23 × 33 × 39; 34.0 ų. c-3 · 78 AO / cell · qc.rs 155 ms; PySCF 5.16 s. 33.3× faster. Grid 23 × 33 × 39; 34.0 ų. lih-8 · 88 AO / cell · qc.rs 1.12 s; PySCF 66.7 s. 59.6× faster. Grid 51 × 45 × 45; 136.1 ų. lih-8 · 88 AO / cell · qc.rs 1.12 s; PySCF 66.7 s. 59.6× faster. Grid 51 × 45 × 45; 136.1 ų. lif-5 · 95 AO / cell · qc.rs 640 ms; PySCF 44.6 s. 69.7× faster. Grid 37 × 43 × 43; 81.2 ų. lif-5 · 95 AO / cell · qc.rs 640 ms; PySCF 44.6 s. 69.7× faster. Grid 37 × 43 × 43; 81.2 ų. lih-9 · 99 AO / cell · qc.rs 1.26 s; PySCF 77.3 s. 61.2× faster. Grid 45 × 45 × 57; 153.2 ų. lih-9 · 99 AO / cell · qc.rs 1.26 s; PySCF 77.3 s. 61.2× faster. Grid 45 × 45 × 57; 153.2 ų. c-4 · 104 AO / cell · qc.rs 186 ms; PySCF 5.36 s. 28.9× faster. Grid 33 × 33 × 33; 45.4 ų. c-4 · 104 AO / cell · qc.rs 186 ms; PySCF 5.36 s. 28.9× faster. Grid 33 × 33 × 33; 45.4 ų. lih-10 · 110 AO / cell · qc.rs 1.39 s; PySCF 83.6 s. 60.1× faster. Grid 45 × 57 × 51; 170.2 ų. lih-10 · 110 AO / cell · qc.rs 1.39 s; PySCF 83.6 s. 60.1× faster. Grid 45 × 57 × 51; 170.2 ų. lif-6 · 114 AO / cell · qc.rs 1.04 s; PySCF 54.2 s. 51.9× faster. Grid 43 × 43 × 43; 97.4 ų. lif-6 · 114 AO / cell · qc.rs 1.04 s; PySCF 54.2 s. 51.9× faster. Grid 43 × 43 × 43; 97.4 ų. c-5 · 130 AO / cell · qc.rs 358 ms; PySCF 8.47 s. 23.7× faster. Grid 39 × 33 × 39; 56.7 ų. c-5 · 130 AO / cell · qc.rs 358 ms; PySCF 8.47 s. 23.7× faster. Grid 39 × 33 × 39; 56.7 ų. lif-7 · 133 AO / cell · qc.rs 930 ms; PySCF 54.9 s. 59.1× faster. Grid 43 × 43 × 43; 113.7 ų. lif-7 · 133 AO / cell · qc.rs 930 ms; PySCF 54.9 s. 59.1× faster. Grid 43 × 43 × 43; 113.7 ų. lif-8 · 152 AO / cell · qc.rs 1.18 s; PySCF 34.9 s. 29.5× faster. Grid 51 × 43 × 43; 129.9 ų. lif-8 · 152 AO / cell · qc.rs 1.18 s; PySCF 34.9 s. 29.5× faster. Grid 51 × 43 × 43; 129.9 ų. c-6 · 156 AO / cell · qc.rs 461 ms; PySCF 10.8 s. 23.4× faster. Grid 39 × 39 × 39; 68.1 ų. c-6 · 156 AO / cell · qc.rs 461 ms; PySCF 10.8 s. 23.4× faster. Grid 39 × 39 × 39; 68.1 ų. c-7 · 182 AO / cell · qc.rs 651 ms; PySCF 10.5 s. 16.1× faster. Grid 39 × 39 × 39; 79.4 ų. c-7 · 182 AO / cell · qc.rs 651 ms; PySCF 10.5 s. 16.1× faster. Grid 39 × 39 × 39; 79.4 ų. c-8 · 208 AO / cell · qc.rs 800 ms; PySCF 12.5 s. 15.6× faster. Grid 45 × 39 × 39; 90.8 ų. c-8 · 208 AO / cell · qc.rs 800 ms; PySCF 12.5 s. 15.6× faster. Grid 45 × 39 × 39; 90.8 ų. c-9 · 234 AO / cell · qc.rs 1.04 s; PySCF 14.3 s. 13.8× faster. Grid 39 × 39 × 49; 102.1 ų. c-9 · 234 AO / cell · qc.rs 1.04 s; PySCF 14.3 s. 13.8× faster. Grid 39 × 39 × 49; 102.1 ų. c-10 · 260 AO / cell · qc.rs 1.45 s; PySCF 16.2 s. 11.2× faster. Grid 39 × 49 × 45; 113.5 ų. c-10 · 260 AO / cell · qc.rs 1.45 s; PySCF 16.2 s. 11.2× faster. Grid 39 × 49 × 45; 113.5 ų. Peak host RSS 95.4 MiB 954 MiB 9.31 GiB 10 20 50 100 200 AO per cell lih-1 · 11 AO / cell · qc.rs 145 MiB; PySCF 162 MiB. 11% less memory. Grid 27 × 27 × 27; 17.0 ų. lih-1 · 11 AO / cell · qc.rs 145 MiB; PySCF 162 MiB. 11% less memory. Grid 27 × 27 × 27; 17.0 ų. lif-1 · 19 AO / cell · qc.rs 167 MiB; PySCF 186 MiB. 10% less memory. Grid 27 × 27 × 27; 16.2 ų. lif-1 · 19 AO / cell · qc.rs 167 MiB; PySCF 186 MiB. 10% less memory. Grid 27 × 27 × 27; 16.2 ų. lih-2 · 22 AO / cell · qc.rs 187 MiB; PySCF 213 MiB. 12% less memory. Grid 27 × 37 × 27; 34.0 ų. lih-2 · 22 AO / cell · qc.rs 187 MiB; PySCF 213 MiB. 12% less memory. Grid 27 × 37 × 27; 34.0 ų. oli2-1 · 25 AO / cell · qc.rs 192 MiB; PySCF 227 MiB. 15% less memory. Grid 29 × 29 × 29; 24.7 ų. oli2-1 · 25 AO / cell · qc.rs 192 MiB; PySCF 227 MiB. 15% less memory. Grid 29 × 29 × 29; 24.7 ų. c-1 · 26 AO / cell · qc.rs 118 MiB; PySCF 168 MiB. 29% less memory. Grid 23 × 23 × 23; 11.3 ų. c-1 · 26 AO / cell · qc.rs 118 MiB; PySCF 168 MiB. 29% less memory. Grid 23 × 23 × 23; 11.3 ų. kf-1 · 27 AO / cell · qc.rs 191 MiB; PySCF 328 MiB. 42% less memory. Grid 35 × 35 × 35; 38.5 ų. kf-1 · 27 AO / cell · qc.rs 191 MiB; PySCF 328 MiB. 42% less memory. Grid 35 × 35 × 35; 38.5 ų. lih-3 · 33 AO / cell · qc.rs 264 MiB; PySCF 347 MiB. 24% less memory. Grid 27 × 37 × 45; 51.1 ų. lih-3 · 33 AO / cell · qc.rs 264 MiB; PySCF 347 MiB. 24% less memory. Grid 27 × 37 × 45; 51.1 ų. lif-2 · 38 AO / cell · qc.rs 200 MiB; PySCF 302 MiB. 34% less memory. Grid 27 × 37 × 27; 32.5 ų. lif-2 · 38 AO / cell · qc.rs 200 MiB; PySCF 302 MiB. 34% less memory. Grid 27 × 37 × 27; 32.5 ų. lih-4 · 44 AO / cell · qc.rs 287 MiB; PySCF 465 MiB. 38% less memory. Grid 37 × 37 × 37; 68.1 ų. lih-4 · 44 AO / cell · qc.rs 287 MiB; PySCF 465 MiB. 38% less memory. Grid 37 × 37 × 37; 68.1 ų. c-2 · 52 AO / cell · qc.rs 173 MiB; PySCF 271 MiB. 36% less memory. Grid 23 × 33 × 23; 22.7 ų. c-2 · 52 AO / cell · qc.rs 173 MiB; PySCF 271 MiB. 36% less memory. Grid 23 × 33 × 23; 22.7 ų. lih-5 · 55 AO / cell · qc.rs 427 MiB; PySCF 839 MiB. 49% less memory. Grid 37 × 45 × 45; 85.1 ų. lih-5 · 55 AO / cell · qc.rs 427 MiB; PySCF 839 MiB. 49% less memory. Grid 37 × 45 × 45; 85.1 ų. lif-3 · 57 AO / cell · qc.rs 268 MiB; PySCF 562 MiB. 52% less memory. Grid 27 × 37 × 43; 48.7 ų. lif-3 · 57 AO / cell · qc.rs 268 MiB; PySCF 562 MiB. 52% less memory. Grid 27 × 37 × 43; 48.7 ų. lih-6 · 66 AO / cell · qc.rs 548 MiB; PySCF 1.06 GiB. 49% less memory. Grid 45 × 45 × 45; 102.1 ų. lih-6 · 66 AO / cell · qc.rs 548 MiB; PySCF 1.06 GiB. 49% less memory. Grid 45 × 45 × 45; 102.1 ų. lif-4 · 76 AO / cell · qc.rs 350 MiB; PySCF 813 MiB. 57% less memory. Grid 37 × 37 × 37; 65.0 ų. lif-4 · 76 AO / cell · qc.rs 350 MiB; PySCF 813 MiB. 57% less memory. Grid 37 × 37 × 37; 65.0 ų. lih-7 · 77 AO / cell · qc.rs 616 MiB; PySCF 1.18 GiB. 49% less memory. Grid 45 × 45 × 45; 119.1 ų. lih-7 · 77 AO / cell · qc.rs 616 MiB; PySCF 1.18 GiB. 49% less memory. Grid 45 × 45 × 45; 119.1 ų. c-3 · 78 AO / cell · qc.rs 259 MiB; PySCF 472 MiB. 45% less memory. Grid 23 × 33 × 39; 34.0 ų. c-3 · 78 AO / cell · qc.rs 259 MiB; PySCF 472 MiB. 45% less memory. Grid 23 × 33 × 39; 34.0 ų. lih-8 · 88 AO / cell · qc.rs 660 MiB; PySCF 1.53 GiB. 58% less memory. Grid 51 × 45 × 45; 136.1 ų. lih-8 · 88 AO / cell · qc.rs 660 MiB; PySCF 1.53 GiB. 58% less memory. Grid 51 × 45 × 45; 136.1 ų. lif-5 · 95 AO / cell · qc.rs 466 MiB; PySCF 1.22 GiB. 63% less memory. Grid 37 × 43 × 43; 81.2 ų. lif-5 · 95 AO / cell · qc.rs 466 MiB; PySCF 1.22 GiB. 63% less memory. Grid 37 × 43 × 43; 81.2 ų. lih-9 · 99 AO / cell · qc.rs 792 MiB; PySCF 1.72 GiB. 55% less memory. Grid 45 × 45 × 57; 153.2 ų. lih-9 · 99 AO / cell · qc.rs 792 MiB; PySCF 1.72 GiB. 55% less memory. Grid 45 × 45 × 57; 153.2 ų. c-4 · 104 AO / cell · qc.rs 327 MiB; PySCF 723 MiB. 55% less memory. Grid 33 × 33 × 33; 45.4 ų. c-4 · 104 AO / cell · qc.rs 327 MiB; PySCF 723 MiB. 55% less memory. Grid 33 × 33 × 33; 45.4 ų. lih-10 · 110 AO / cell · qc.rs 958 MiB; PySCF 2.03 GiB. 54% less memory. Grid 45 × 57 × 51; 170.2 ų. lih-10 · 110 AO / cell · qc.rs 958 MiB; PySCF 2.03 GiB. 54% less memory. Grid 45 × 57 × 51; 170.2 ų. lif-6 · 114 AO / cell · qc.rs 662 MiB; PySCF 1.56 GiB. 59% less memory. Grid 43 × 43 × 43; 97.4 ų. lif-6 · 114 AO / cell · qc.rs 662 MiB; PySCF 1.56 GiB. 59% less memory. Grid 43 × 43 × 43; 97.4 ų. c-5 · 130 AO / cell · qc.rs 477 MiB; PySCF 1.14 GiB. 59% less memory. Grid 39 × 33 × 39; 56.7 ų. c-5 · 130 AO / cell · qc.rs 477 MiB; PySCF 1.14 GiB. 59% less memory. Grid 39 × 33 × 39; 56.7 ų. lif-7 · 133 AO / cell · qc.rs 741 MiB; PySCF 1.81 GiB. 60% less memory. Grid 43 × 43 × 43; 113.7 ų. lif-7 · 133 AO / cell · qc.rs 741 MiB; PySCF 1.81 GiB. 60% less memory. Grid 43 × 43 × 43; 113.7 ų. lif-8 · 152 AO / cell · qc.rs 942 MiB; PySCF 2.38 GiB. 61% less memory. Grid 51 × 43 × 43; 129.9 ų. lif-8 · 152 AO / cell · qc.rs 942 MiB; PySCF 2.38 GiB. 61% less memory. Grid 51 × 43 × 43; 129.9 ų. c-6 · 156 AO / cell · qc.rs 622 MiB; PySCF 1.49 GiB. 59% less memory. Grid 39 × 39 × 39; 68.1 ų. c-6 · 156 AO / cell · qc.rs 622 MiB; PySCF 1.49 GiB. 59% less memory. Grid 39 × 39 × 39; 68.1 ų. c-7 · 182 AO / cell · qc.rs 700 MiB; PySCF 1.81 GiB. 62% less memory. Grid 39 × 39 × 39; 79.4 ų. c-7 · 182 AO / cell · qc.rs 700 MiB; PySCF 1.81 GiB. 62% less memory. Grid 39 × 39 × 39; 79.4 ų. c-8 · 208 AO / cell · qc.rs 890 MiB; PySCF 2.43 GiB. 64% less memory. Grid 45 × 39 × 39; 90.8 ų. c-8 · 208 AO / cell · qc.rs 890 MiB; PySCF 2.43 GiB. 64% less memory. Grid 45 × 39 × 39; 90.8 ų. c-9 · 234 AO / cell · qc.rs 1.03 GiB; PySCF 3.32 GiB. 69% less memory. Grid 39 × 39 × 49; 102.1 ų. c-9 · 234 AO / cell · qc.rs 1.03 GiB; PySCF 3.32 GiB. 69% less memory. Grid 39 × 39 × 49; 102.1 ų. c-10 · 260 AO / cell · qc.rs 1.29 GiB; PySCF 3.70 GiB. 65% less memory. Grid 39 × 49 × 45; 113.5 ų. c-10 · 260 AO / cell · qc.rs 1.29 GiB; PySCF 3.70 GiB. 65% less memory. Grid 39 × 49 × 45; 113.5 ų.
Apple M3 Ultra Per SCF time 10.0 ms 100 ms 1.00 s 10.0 s 100 s 1.00e+3 s 10 20 50 100 200 AO per cell lih-1 · 11 AO / cell · qc.rs 356 ms; PySCF 12.2 s. 34.3× faster. Grid 27 × 27 × 27; 17.0 ų. lih-1 · 11 AO / cell · qc.rs 356 ms; PySCF 12.2 s. 34.3× faster. Grid 27 × 27 × 27; 17.0 ų. lif-1 · 19 AO / cell · qc.rs 206 ms; PySCF 12.7 s. 61.9× faster. Grid 27 × 27 × 27; 16.2 ų. lif-1 · 19 AO / cell · qc.rs 206 ms; PySCF 12.7 s. 61.9× faster. Grid 27 × 27 × 27; 16.2 ų. lih-2 · 22 AO / cell · qc.rs 235 ms; PySCF 16.8 s. 71.7× faster. Grid 27 × 37 × 27; 34.0 ų. lih-2 · 22 AO / cell · qc.rs 235 ms; PySCF 16.8 s. 71.7× faster. Grid 27 × 37 × 27; 34.0 ų. oli2-1 · 25 AO / cell · qc.rs 274 ms; PySCF 18.8 s. 68.6× faster. Grid 29 × 29 × 29; 24.7 ų. oli2-1 · 25 AO / cell · qc.rs 274 ms; PySCF 18.8 s. 68.6× faster. Grid 29 × 29 × 29; 24.7 ų. c-1 · 26 AO / cell · qc.rs 61.4 ms; PySCF 2.54 s. 41.4× faster. Grid 23 × 23 × 23; 11.3 ų. c-1 · 26 AO / cell · qc.rs 61.4 ms; PySCF 2.54 s. 41.4× faster. Grid 23 × 23 × 23; 11.3 ų. kf-1 · 27 AO / cell · qc.rs 336 ms; PySCF 18.6 s. 55.4× faster. Grid 35 × 35 × 35; 38.5 ų. kf-1 · 27 AO / cell · qc.rs 336 ms; PySCF 18.6 s. 55.4× faster. Grid 35 × 35 × 35; 38.5 ų. lih-3 · 33 AO / cell · qc.rs 487 ms; PySCF 30.6 s. 62.8× faster. Grid 27 × 37 × 45; 51.1 ų. lih-3 · 33 AO / cell · qc.rs 487 ms; PySCF 30.6 s. 62.8× faster. Grid 27 × 37 × 45; 51.1 ų. lif-2 · 38 AO / cell · qc.rs 253 ms; PySCF 17.5 s. 69.2× faster. Grid 27 × 37 × 27; 32.5 ų. lif-2 · 38 AO / cell · qc.rs 253 ms; PySCF 17.5 s. 69.2× faster. Grid 27 × 37 × 27; 32.5 ų. lih-4 · 44 AO / cell · qc.rs 483 ms; PySCF 32.1 s. 66.4× faster. Grid 37 × 37 × 37; 68.1 ų. lih-4 · 44 AO / cell · qc.rs 483 ms; PySCF 32.1 s. 66.4× faster. Grid 37 × 37 × 37; 68.1 ų. c-2 · 52 AO / cell · qc.rs 79.2 ms; PySCF 2.72 s. 34.4× faster. Grid 23 × 33 × 23; 22.7 ų. c-2 · 52 AO / cell · qc.rs 79.2 ms; PySCF 2.72 s. 34.4× faster. Grid 23 × 33 × 23; 22.7 ų. lih-5 · 55 AO / cell · qc.rs 825 ms; PySCF 54.1 s. 65.6× faster. Grid 37 × 45 × 45; 85.1 ų. lih-5 · 55 AO / cell · qc.rs 825 ms; PySCF 54.1 s. 65.6× faster. Grid 37 × 45 × 45; 85.1 ų. lif-3 · 57 AO / cell · qc.rs 521 ms; PySCF 30.9 s. 59.2× faster. Grid 27 × 37 × 43; 48.7 ų. lif-3 · 57 AO / cell · qc.rs 521 ms; PySCF 30.9 s. 59.2× faster. Grid 27 × 37 × 43; 48.7 ų. lih-6 · 66 AO / cell · qc.rs 1.12 s; PySCF 68.6 s. 61.4× faster. Grid 45 × 45 × 45; 102.1 ų. lih-6 · 66 AO / cell · qc.rs 1.12 s; PySCF 68.6 s. 61.4× faster. Grid 45 × 45 × 45; 102.1 ų. lif-4 · 76 AO / cell · qc.rs 475 ms; PySCF 33.7 s. 71× faster. Grid 37 × 37 × 37; 65.0 ų. lif-4 · 76 AO / cell · qc.rs 475 ms; PySCF 33.7 s. 71× faster. Grid 37 × 37 × 37; 65.0 ų. lih-7 · 77 AO / cell · qc.rs 1.12 s; PySCF 69.3 s. 61.7× faster. Grid 45 × 45 × 45; 119.1 ų. lih-7 · 77 AO / cell · qc.rs 1.12 s; PySCF 69.3 s. 61.7× faster. Grid 45 × 45 × 45; 119.1 ų. c-3 · 78 AO / cell · qc.rs 183 ms; PySCF 5.84 s. 31.9× faster. Grid 23 × 33 × 39; 34.0 ų. c-3 · 78 AO / cell · qc.rs 183 ms; PySCF 5.84 s. 31.9× faster. Grid 23 × 33 × 39; 34.0 ų. lih-8 · 88 AO / cell · qc.rs 1.34 s; PySCF 80.0 s. 59.9× faster. Grid 51 × 45 × 45; 136.1 ų. lih-8 · 88 AO / cell · qc.rs 1.34 s; PySCF 80.0 s. 59.9× faster. Grid 51 × 45 × 45; 136.1 ų. lif-5 · 95 AO / cell · qc.rs 770 ms; PySCF 53.1 s. 69× faster. Grid 37 × 43 × 43; 81.2 ų. lif-5 · 95 AO / cell · qc.rs 770 ms; PySCF 53.1 s. 69× faster. Grid 37 × 43 × 43; 81.2 ų. lih-9 · 99 AO / cell · qc.rs 1.50 s; PySCF 92.1 s. 61.3× faster. Grid 45 × 45 × 57; 153.2 ų. lih-9 · 99 AO / cell · qc.rs 1.50 s; PySCF 92.1 s. 61.3× faster. Grid 45 × 45 × 57; 153.2 ų. c-4 · 104 AO / cell · qc.rs 220 ms; PySCF 6.09 s. 27.6× faster. Grid 33 × 33 × 33; 45.4 ų. c-4 · 104 AO / cell · qc.rs 220 ms; PySCF 6.09 s. 27.6× faster. Grid 33 × 33 × 33; 45.4 ų. lih-10 · 110 AO / cell · qc.rs 1.64 s; PySCF 100 s. 61.1× faster. Grid 45 × 57 × 51; 170.2 ų. lih-10 · 110 AO / cell · qc.rs 1.64 s; PySCF 100 s. 61.1× faster. Grid 45 × 57 × 51; 170.2 ų. lif-6 · 114 AO / cell · qc.rs 1.26 s; PySCF 65.2 s. 51.6× faster. Grid 43 × 43 × 43; 97.4 ų. lif-6 · 114 AO / cell · qc.rs 1.26 s; PySCF 65.2 s. 51.6× faster. Grid 43 × 43 × 43; 97.4 ų. c-5 · 130 AO / cell · qc.rs 427 ms; PySCF 9.71 s. 22.7× faster. Grid 39 × 33 × 39; 56.7 ų. c-5 · 130 AO / cell · qc.rs 427 ms; PySCF 9.71 s. 22.7× faster. Grid 39 × 33 × 39; 56.7 ų. lif-7 · 133 AO / cell · qc.rs 1.12 s; PySCF 66.2 s. 59× faster. Grid 43 × 43 × 43; 113.7 ų. lif-7 · 133 AO / cell · qc.rs 1.12 s; PySCF 66.2 s. 59× faster. Grid 43 × 43 × 43; 113.7 ų. lif-8 · 152 AO / cell · qc.rs 1.43 s; PySCF 42.0 s. 29.4× faster. Grid 51 × 43 × 43; 129.9 ų. lif-8 · 152 AO / cell · qc.rs 1.43 s; PySCF 42.0 s. 29.4× faster. Grid 51 × 43 × 43; 129.9 ų. c-6 · 156 AO / cell · qc.rs 548 ms; PySCF 12.5 s. 22.9× faster. Grid 39 × 39 × 39; 68.1 ų. c-6 · 156 AO / cell · qc.rs 548 ms; PySCF 12.5 s. 22.9× faster. Grid 39 × 39 × 39; 68.1 ų. c-7 · 182 AO / cell · qc.rs 774 ms; PySCF 12.0 s. 15.5× faster. Grid 39 × 39 × 39; 79.4 ų. c-7 · 182 AO / cell · qc.rs 774 ms; PySCF 12.0 s. 15.5× faster. Grid 39 × 39 × 39; 79.4 ų. c-8 · 208 AO / cell · qc.rs 936 ms; PySCF 14.5 s. 15.5× faster. Grid 45 × 39 × 39; 90.8 ų. c-8 · 208 AO / cell · qc.rs 936 ms; PySCF 14.5 s. 15.5× faster. Grid 45 × 39 × 39; 90.8 ų. c-9 · 234 AO / cell · qc.rs 1.25 s; PySCF 16.5 s. 13.2× faster. Grid 39 × 39 × 49; 102.1 ų. c-9 · 234 AO / cell · qc.rs 1.25 s; PySCF 16.5 s. 13.2× faster. Grid 39 × 39 × 49; 102.1 ų. c-10 · 260 AO / cell · qc.rs 1.73 s; PySCF 18.7 s. 10.8× faster. Grid 39 × 49 × 45; 113.5 ų. c-10 · 260 AO / cell · qc.rs 1.73 s; PySCF 18.7 s. 10.8× faster. Grid 39 × 49 × 45; 113.5 ų. Peak host RSS 95.4 MiB 954 MiB 9.31 GiB 10 20 50 100 200 AO per cell lih-1 · 11 AO / cell · qc.rs 147 MiB; PySCF 216 MiB. 32% less memory. Grid 27 × 27 × 27; 17.0 ų. lih-1 · 11 AO / cell · qc.rs 147 MiB; PySCF 216 MiB. 32% less memory. Grid 27 × 27 × 27; 17.0 ų. lif-1 · 19 AO / cell · qc.rs 195 MiB; PySCF 324 MiB. 40% less memory. Grid 27 × 27 × 27; 16.2 ų. lif-1 · 19 AO / cell · qc.rs 195 MiB; PySCF 324 MiB. 40% less memory. Grid 27 × 27 × 27; 16.2 ų. lih-2 · 22 AO / cell · qc.rs 210 MiB; PySCF 354 MiB. 41% less memory. Grid 27 × 37 × 27; 34.0 ų. lih-2 · 22 AO / cell · qc.rs 210 MiB; PySCF 354 MiB. 41% less memory. Grid 27 × 37 × 27; 34.0 ų. oli2-1 · 25 AO / cell · qc.rs 262 MiB; PySCF 433 MiB. 40% less memory. Grid 29 × 29 × 29; 24.7 ų. oli2-1 · 25 AO / cell · qc.rs 262 MiB; PySCF 433 MiB. 40% less memory. Grid 29 × 29 × 29; 24.7 ų. c-1 · 26 AO / cell · qc.rs 129 MiB; PySCF 174 MiB. 26% less memory. Grid 23 × 23 × 23; 11.3 ų. c-1 · 26 AO / cell · qc.rs 129 MiB; PySCF 174 MiB. 26% less memory. Grid 23 × 23 × 23; 11.3 ų. kf-1 · 27 AO / cell · qc.rs 315 MiB; PySCF 490 MiB. 36% less memory. Grid 35 × 35 × 35; 38.5 ų. kf-1 · 27 AO / cell · qc.rs 315 MiB; PySCF 490 MiB. 36% less memory. Grid 35 × 35 × 35; 38.5 ų. lih-3 · 33 AO / cell · qc.rs 391 MiB; PySCF 473 MiB. 17% less memory. Grid 27 × 37 × 45; 51.1 ų. lih-3 · 33 AO / cell · qc.rs 391 MiB; PySCF 473 MiB. 17% less memory. Grid 27 × 37 × 45; 51.1 ų. lif-2 · 38 AO / cell · qc.rs 258 MiB; PySCF 737 MiB. 65% less memory. Grid 27 × 37 × 27; 32.5 ų. lif-2 · 38 AO / cell · qc.rs 258 MiB; PySCF 737 MiB. 65% less memory. Grid 27 × 37 × 27; 32.5 ų. lih-4 · 44 AO / cell · qc.rs 478 MiB; PySCF 691 MiB. 31% less memory. Grid 37 × 37 × 37; 68.1 ų. lih-4 · 44 AO / cell · qc.rs 478 MiB; PySCF 691 MiB. 31% less memory. Grid 37 × 37 × 37; 68.1 ų. c-2 · 52 AO / cell · qc.rs 204 MiB; PySCF 511 MiB. 60% less memory. Grid 23 × 33 × 23; 22.7 ų. c-2 · 52 AO / cell · qc.rs 204 MiB; PySCF 511 MiB. 60% less memory. Grid 23 × 33 × 23; 22.7 ų. lih-5 · 55 AO / cell · qc.rs 792 MiB; PySCF 1.05 GiB. 26% less memory. Grid 37 × 45 × 45; 85.1 ų. lih-5 · 55 AO / cell · qc.rs 792 MiB; PySCF 1.05 GiB. 26% less memory. Grid 37 × 45 × 45; 85.1 ų. lif-3 · 57 AO / cell · qc.rs 475 MiB; PySCF 801 MiB. 41% less memory. Grid 27 × 37 × 43; 48.7 ų. lif-3 · 57 AO / cell · qc.rs 475 MiB; PySCF 801 MiB. 41% less memory. Grid 27 × 37 × 43; 48.7 ų. lih-6 · 66 AO / cell · qc.rs 1.02e+3 MiB; PySCF 1.32 GiB. 25% less memory. Grid 45 × 45 × 45; 102.1 ų. lih-6 · 66 AO / cell · qc.rs 1.02e+3 MiB; PySCF 1.32 GiB. 25% less memory. Grid 45 × 45 × 45; 102.1 ų. lif-4 · 76 AO / cell · qc.rs 703 MiB; PySCF 1.01e+3 MiB. 31% less memory. Grid 37 × 37 × 37; 65.0 ų. lif-4 · 76 AO / cell · qc.rs 703 MiB; PySCF 1.01e+3 MiB. 31% less memory. Grid 37 × 37 × 37; 65.0 ų. lih-7 · 77 AO / cell · qc.rs 1.04 GiB; PySCF 1.42 GiB. 27% less memory. Grid 45 × 45 × 45; 119.1 ų. lih-7 · 77 AO / cell · qc.rs 1.04 GiB; PySCF 1.42 GiB. 27% less memory. Grid 45 × 45 × 45; 119.1 ų. c-3 · 78 AO / cell · qc.rs 373 MiB; PySCF 541 MiB. 31% less memory. Grid 23 × 33 × 39; 34.0 ų. c-3 · 78 AO / cell · qc.rs 373 MiB; PySCF 541 MiB. 31% less memory. Grid 23 × 33 × 39; 34.0 ų. lih-8 · 88 AO / cell · qc.rs 1.17 GiB; PySCF 1.83 GiB. 36% less memory. Grid 51 × 45 × 45; 136.1 ų. lih-8 · 88 AO / cell · qc.rs 1.17 GiB; PySCF 1.83 GiB. 36% less memory. Grid 51 × 45 × 45; 136.1 ų. lif-5 · 95 AO / cell · qc.rs 969 MiB; PySCF 1.47 GiB. 36% less memory. Grid 37 × 43 × 43; 81.2 ų. lif-5 · 95 AO / cell · qc.rs 969 MiB; PySCF 1.47 GiB. 36% less memory. Grid 37 × 43 × 43; 81.2 ų. lih-9 · 99 AO / cell · qc.rs 1.40 GiB; PySCF 2.06 GiB. 32% less memory. Grid 45 × 45 × 57; 153.2 ų. lih-9 · 99 AO / cell · qc.rs 1.40 GiB; PySCF 2.06 GiB. 32% less memory. Grid 45 × 45 × 57; 153.2 ų. c-4 · 104 AO / cell · qc.rs 398 MiB; PySCF 889 MiB. 55% less memory. Grid 33 × 33 × 33; 45.4 ų. c-4 · 104 AO / cell · qc.rs 398 MiB; PySCF 889 MiB. 55% less memory. Grid 33 × 33 × 33; 45.4 ų. lih-10 · 110 AO / cell · qc.rs 1.71 GiB; PySCF 2.41 GiB. 29% less memory. Grid 45 × 57 × 51; 170.2 ų. lih-10 · 110 AO / cell · qc.rs 1.71 GiB; PySCF 2.41 GiB. 29% less memory. Grid 45 × 57 × 51; 170.2 ų. lif-6 · 114 AO / cell · qc.rs 1.15 GiB; PySCF 1.81 GiB. 37% less memory. Grid 43 × 43 × 43; 97.4 ų. lif-6 · 114 AO / cell · qc.rs 1.15 GiB; PySCF 1.81 GiB. 37% less memory. Grid 43 × 43 × 43; 97.4 ų. c-5 · 130 AO / cell · qc.rs 869 MiB; PySCF 1.42 GiB. 40% less memory. Grid 39 × 33 × 39; 56.7 ų. c-5 · 130 AO / cell · qc.rs 869 MiB; PySCF 1.42 GiB. 40% less memory. Grid 39 × 33 × 39; 56.7 ų. lif-7 · 133 AO / cell · qc.rs 1.41 GiB; PySCF 1.99 GiB. 29% less memory. Grid 43 × 43 × 43; 113.7 ų. lif-7 · 133 AO / cell · qc.rs 1.41 GiB; PySCF 1.99 GiB. 29% less memory. Grid 43 × 43 × 43; 113.7 ų. lif-8 · 152 AO / cell · qc.rs 1.75 GiB; PySCF 2.69 GiB. 35% less memory. Grid 51 × 43 × 43; 129.9 ų. lif-8 · 152 AO / cell · qc.rs 1.75 GiB; PySCF 2.69 GiB. 35% less memory. Grid 51 × 43 × 43; 129.9 ų. c-6 · 156 AO / cell · qc.rs 1.31 GiB; PySCF 1.67 GiB. 22% less memory. Grid 39 × 39 × 39; 68.1 ų. c-6 · 156 AO / cell · qc.rs 1.31 GiB; PySCF 1.67 GiB. 22% less memory. Grid 39 × 39 × 39; 68.1 ų. c-7 · 182 AO / cell · qc.rs 1.50 GiB; PySCF 1.97 GiB. 24% less memory. Grid 39 × 39 × 39; 79.4 ų. c-7 · 182 AO / cell · qc.rs 1.50 GiB; PySCF 1.97 GiB. 24% less memory. Grid 39 × 39 × 39; 79.4 ų. c-8 · 208 AO / cell · qc.rs 1.85 GiB; PySCF 2.44 GiB. 24% less memory. Grid 45 × 39 × 39; 90.8 ų. c-8 · 208 AO / cell · qc.rs 1.85 GiB; PySCF 2.44 GiB. 24% less memory. Grid 45 × 39 × 39; 90.8 ų. c-9 · 234 AO / cell · qc.rs 2.18 GiB; PySCF 3.07 GiB. 29% less memory. Grid 39 × 39 × 49; 102.1 ų. c-9 · 234 AO / cell · qc.rs 2.18 GiB; PySCF 3.07 GiB. 29% less memory. Grid 39 × 39 × 49; 102.1 ų. c-10 · 260 AO / cell · qc.rs 2.39 GiB; PySCF 3.25 GiB. 26% less memory. Grid 39 × 49 × 45; 113.5 ų. c-10 · 260 AO / cell · qc.rs 2.39 GiB; PySCF 3.25 GiB. 26% less memory. Grid 39 × 49 × 45; 113.5 ų.
AMD EPYC 7702P Per SCF time 10.0 ms 100 ms 1.00 s 10.0 s 100 s 1.00e+3 s 10 20 50 100 200 AO per cell lih-1 · 11 AO / cell · qc.rs 1.11 s; PySCF 7.08 s. 6.4× faster. Grid 27 × 27 × 27; 17.0 ų. lih-1 · 11 AO / cell · qc.rs 1.11 s; PySCF 7.08 s. 6.4× faster. Grid 27 × 27 × 27; 17.0 ų. lif-1 · 19 AO / cell · qc.rs 584 ms; PySCF 7.14 s. 12.2× faster. Grid 27 × 27 × 27; 16.2 ų. lif-1 · 19 AO / cell · qc.rs 584 ms; PySCF 7.14 s. 12.2× faster. Grid 27 × 27 × 27; 16.2 ų. lih-2 · 22 AO / cell · qc.rs 730 ms; PySCF 9.21 s. 12.6× faster. Grid 27 × 37 × 27; 34.0 ų. lih-2 · 22 AO / cell · qc.rs 730 ms; PySCF 9.21 s. 12.6× faster. Grid 27 × 37 × 27; 34.0 ų. c-1 · 26 AO / cell · qc.rs 172 ms; PySCF 1.27 s. 7.4× faster. Grid 23 × 23 × 23; 11.3 ų. c-1 · 26 AO / cell · qc.rs 172 ms; PySCF 1.27 s. 7.4× faster. Grid 23 × 23 × 23; 11.3 ų. lih-3 · 33 AO / cell · qc.rs 1.59 s; PySCF 16.6 s. 10.5× faster. Grid 27 × 37 × 45; 51.1 ų. lih-3 · 33 AO / cell · qc.rs 1.59 s; PySCF 16.6 s. 10.5× faster. Grid 27 × 37 × 45; 51.1 ų. lif-2 · 38 AO / cell · qc.rs 745 ms; PySCF 8.94 s. 12× faster. Grid 27 × 37 × 27; 32.5 ų. lif-2 · 38 AO / cell · qc.rs 745 ms; PySCF 8.94 s. 12× faster. Grid 27 × 37 × 27; 32.5 ų. lih-4 · 44 AO / cell · qc.rs 1.58 s; PySCF 17.1 s. 10.8× faster. Grid 37 × 37 × 37; 68.1 ų. lih-4 · 44 AO / cell · qc.rs 1.58 s; PySCF 17.1 s. 10.8× faster. Grid 37 × 37 × 37; 68.1 ų. c-2 · 52 AO / cell · qc.rs 236 ms; PySCF 1.31 s. 5.6× faster. Grid 23 × 33 × 23; 22.7 ų. c-2 · 52 AO / cell · qc.rs 236 ms; PySCF 1.31 s. 5.6× faster. Grid 23 × 33 × 23; 22.7 ų. lih-5 · 55 AO / cell · qc.rs 2.52 s; PySCF 28.0 s. 11.1× faster. Grid 37 × 45 × 45; 85.1 ų. lih-5 · 55 AO / cell · qc.rs 2.52 s; PySCF 28.0 s. 11.1× faster. Grid 37 × 45 × 45; 85.1 ų. lif-3 · 57 AO / cell · qc.rs 1.53 s; PySCF 16.2 s. 10.6× faster. Grid 27 × 37 × 43; 48.7 ų. lif-3 · 57 AO / cell · qc.rs 1.53 s; PySCF 16.2 s. 10.6× faster. Grid 27 × 37 × 43; 48.7 ų. lih-6 · 66 AO / cell · qc.rs 3.35 s; PySCF 34.8 s. 10.4× faster. Grid 45 × 45 × 45; 102.1 ų. lih-6 · 66 AO / cell · qc.rs 3.35 s; PySCF 34.8 s. 10.4× faster. Grid 45 × 45 × 45; 102.1 ų. lif-4 · 76 AO / cell · qc.rs 1.40 s; PySCF 17.3 s. 12.4× faster. Grid 37 × 37 × 37; 65.0 ų. lif-4 · 76 AO / cell · qc.rs 1.40 s; PySCF 17.3 s. 12.4× faster. Grid 37 × 37 × 37; 65.0 ų. lih-7 · 77 AO / cell · qc.rs 3.31 s; PySCF 35.1 s. 10.6× faster. Grid 45 × 45 × 45; 119.1 ų. lih-7 · 77 AO / cell · qc.rs 3.31 s; PySCF 35.1 s. 10.6× faster. Grid 45 × 45 × 45; 119.1 ų. c-3 · 78 AO / cell · qc.rs 548 ms; PySCF 2.87 s. 5.2× faster. Grid 23 × 33 × 39; 34.0 ų. c-3 · 78 AO / cell · qc.rs 548 ms; PySCF 2.87 s. 5.2× faster. Grid 23 × 33 × 39; 34.0 ų. lih-8 · 88 AO / cell · qc.rs 3.88 s; PySCF 39.8 s. 10.3× faster. Grid 51 × 45 × 45; 136.1 ų. lih-8 · 88 AO / cell · qc.rs 3.88 s; PySCF 39.8 s. 10.3× faster. Grid 51 × 45 × 45; 136.1 ų. lif-5 · 95 AO / cell · qc.rs 2.21 s; PySCF 25.6 s. 11.6× faster. Grid 37 × 43 × 43; 81.2 ų. lif-5 · 95 AO / cell · qc.rs 2.21 s; PySCF 25.6 s. 11.6× faster. Grid 37 × 43 × 43; 81.2 ų. lih-9 · 99 AO / cell · qc.rs 4.47 s; PySCF 47.0 s. 10.5× faster. Grid 45 × 45 × 57; 153.2 ų. lih-9 · 99 AO / cell · qc.rs 4.47 s; PySCF 47.0 s. 10.5× faster. Grid 45 × 45 × 57; 153.2 ų. c-4 · 104 AO / cell · qc.rs 670 ms; PySCF 2.93 s. 4.4× faster. Grid 33 × 33 × 33; 45.4 ų. c-4 · 104 AO / cell · qc.rs 670 ms; PySCF 2.93 s. 4.4× faster. Grid 33 × 33 × 33; 45.4 ų. lih-10 · 110 AO / cell · qc.rs 4.94 s; PySCF 50.8 s. 10.3× faster. Grid 45 × 57 × 51; 170.2 ų. lih-10 · 110 AO / cell · qc.rs 4.94 s; PySCF 50.8 s. 10.3× faster. Grid 45 × 57 × 51; 170.2 ų. lif-6 · 114 AO / cell · qc.rs 3.53 s; PySCF 31.0 s. 8.8× faster. Grid 43 × 43 × 43; 97.4 ų. lif-6 · 114 AO / cell · qc.rs 3.53 s; PySCF 31.0 s. 8.8× faster. Grid 43 × 43 × 43; 97.4 ų. c-5 · 130 AO / cell · qc.rs 1.31 s; PySCF 4.59 s. 3.5× faster. Grid 39 × 33 × 39; 56.7 ų. c-5 · 130 AO / cell · qc.rs 1.31 s; PySCF 4.59 s. 3.5× faster. Grid 39 × 33 × 39; 56.7 ų. lif-7 · 133 AO / cell · qc.rs 3.16 s; PySCF 31.0 s. 9.8× faster. Grid 43 × 43 × 43; 113.7 ų. lif-7 · 133 AO / cell · qc.rs 3.16 s; PySCF 31.0 s. 9.8× faster. Grid 43 × 43 × 43; 113.7 ų. lif-8 · 152 AO / cell · qc.rs 4.00 s; PySCF 19.8 s. 5× faster. Grid 51 × 43 × 43; 129.9 ų. lif-8 · 152 AO / cell · qc.rs 4.00 s; PySCF 19.8 s. 5× faster. Grid 51 × 43 × 43; 129.9 ų. c-6 · 156 AO / cell · qc.rs 1.66 s; PySCF 5.72 s. 3.4× faster. Grid 39 × 39 × 39; 68.1 ų. c-6 · 156 AO / cell · qc.rs 1.66 s; PySCF 5.72 s. 3.4× faster. Grid 39 × 39 × 39; 68.1 ų. lif-9 · 171 AO / cell · qc.rs 4.92 s; PySCF 20.5 s. 4.2× faster. Grid 43 × 43 × 57; 146.2 ų. lif-9 · 171 AO / cell · qc.rs 4.92 s; PySCF 20.5 s. 4.2× faster. Grid 43 × 43 × 57; 146.2 ų. c-7 · 182 AO / cell · qc.rs 2.30 s; PySCF 5.72 s. 2.5× faster. Grid 39 × 39 × 39; 79.4 ų. c-7 · 182 AO / cell · qc.rs 2.30 s; PySCF 5.72 s. 2.5× faster. Grid 39 × 39 × 39; 79.4 ų. lif-10 · 190 AO / cell · qc.rs 7.33 s; PySCF 26.9 s. 3.7× faster. Grid 43 × 57 × 51; 162.4 ų. lif-10 · 190 AO / cell · qc.rs 7.33 s; PySCF 26.9 s. 3.7× faster. Grid 43 × 57 × 51; 162.4 ų. c-8 · 208 AO / cell · qc.rs 2.73 s; PySCF 6.82 s. 2.5× faster. Grid 45 × 39 × 39; 90.8 ų. c-8 · 208 AO / cell · qc.rs 2.73 s; PySCF 6.82 s. 2.5× faster. Grid 45 × 39 × 39; 90.8 ų. c-9 · 234 AO / cell · qc.rs 3.64 s; PySCF 8.05 s. 2.2× faster. Grid 39 × 39 × 49; 102.1 ų. c-9 · 234 AO / cell · qc.rs 3.64 s; PySCF 8.05 s. 2.2× faster. Grid 39 × 39 × 49; 102.1 ų. c-10 · 260 AO / cell · qc.rs 4.92 s; PySCF 8.99 s. 1.8× faster. Grid 39 × 49 × 45; 113.5 ų. c-10 · 260 AO / cell · qc.rs 4.92 s; PySCF 8.99 s. 1.8× faster. Grid 39 × 49 × 45; 113.5 ų. Peak host RSS 95.4 MiB 954 MiB 9.31 GiB 10 20 50 100 200 AO per cell lih-1 · 11 AO / cell · qc.rs 138 MiB; PySCF 178 MiB. 22% less memory. Grid 27 × 27 × 27; 17.0 ų. lih-1 · 11 AO / cell · qc.rs 138 MiB; PySCF 178 MiB. 22% less memory. Grid 27 × 27 × 27; 17.0 ų. lif-1 · 19 AO / cell · qc.rs 152 MiB; PySCF 202 MiB. 25% less memory. Grid 27 × 27 × 27; 16.2 ų. lif-1 · 19 AO / cell · qc.rs 152 MiB; PySCF 202 MiB. 25% less memory. Grid 27 × 27 × 27; 16.2 ų. lih-2 · 22 AO / cell · qc.rs 178 MiB; PySCF 204 MiB. 13% less memory. Grid 27 × 37 × 27; 34.0 ų. lih-2 · 22 AO / cell · qc.rs 178 MiB; PySCF 204 MiB. 13% less memory. Grid 27 × 37 × 27; 34.0 ų. c-1 · 26 AO / cell · qc.rs 126 MiB; PySCF 172 MiB. 27% less memory. Grid 23 × 23 × 23; 11.3 ų. c-1 · 26 AO / cell · qc.rs 126 MiB; PySCF 172 MiB. 27% less memory. Grid 23 × 23 × 23; 11.3 ų. lih-3 · 33 AO / cell · qc.rs 251 MiB; PySCF 317 MiB. 21% less memory. Grid 27 × 37 × 45; 51.1 ų. lih-3 · 33 AO / cell · qc.rs 251 MiB; PySCF 317 MiB. 21% less memory. Grid 27 × 37 × 45; 51.1 ų. lif-2 · 38 AO / cell · qc.rs 193 MiB; PySCF 250 MiB. 23% less memory. Grid 27 × 37 × 27; 32.5 ų. lif-2 · 38 AO / cell · qc.rs 193 MiB; PySCF 250 MiB. 23% less memory. Grid 27 × 37 × 27; 32.5 ų. lih-4 · 44 AO / cell · qc.rs 275 MiB; PySCF 379 MiB. 27% less memory. Grid 37 × 37 × 37; 68.1 ų. lih-4 · 44 AO / cell · qc.rs 275 MiB; PySCF 379 MiB. 27% less memory. Grid 37 × 37 × 37; 68.1 ų. c-2 · 52 AO / cell · qc.rs 177 MiB; PySCF 245 MiB. 28% less memory. Grid 23 × 33 × 23; 22.7 ų. c-2 · 52 AO / cell · qc.rs 177 MiB; PySCF 245 MiB. 28% less memory. Grid 23 × 33 × 23; 22.7 ų. lih-5 · 55 AO / cell · qc.rs 396 MiB; PySCF 537 MiB. 26% less memory. Grid 37 × 45 × 45; 85.1 ų. lih-5 · 55 AO / cell · qc.rs 396 MiB; PySCF 537 MiB. 26% less memory. Grid 37 × 45 × 45; 85.1 ų. lif-3 · 57 AO / cell · qc.rs 268 MiB; PySCF 431 MiB. 38% less memory. Grid 27 × 37 × 43; 48.7 ų. lif-3 · 57 AO / cell · qc.rs 268 MiB; PySCF 431 MiB. 38% less memory. Grid 27 × 37 × 43; 48.7 ų. lih-6 · 66 AO / cell · qc.rs 533 MiB; PySCF 721 MiB. 26% less memory. Grid 45 × 45 × 45; 102.1 ų. lih-6 · 66 AO / cell · qc.rs 533 MiB; PySCF 721 MiB. 26% less memory. Grid 45 × 45 × 45; 102.1 ų. lif-4 · 76 AO / cell · qc.rs 353 MiB; PySCF 503 MiB. 30% less memory. Grid 37 × 37 × 37; 65.0 ų. lif-4 · 76 AO / cell · qc.rs 353 MiB; PySCF 503 MiB. 30% less memory. Grid 37 × 37 × 37; 65.0 ų. lih-7 · 77 AO / cell · qc.rs 603 MiB; PySCF 820 MiB. 26% less memory. Grid 45 × 45 × 45; 119.1 ų. lih-7 · 77 AO / cell · qc.rs 603 MiB; PySCF 820 MiB. 26% less memory. Grid 45 × 45 × 45; 119.1 ų. c-3 · 78 AO / cell · qc.rs 251 MiB; PySCF 410 MiB. 39% less memory. Grid 23 × 33 × 39; 34.0 ų. c-3 · 78 AO / cell · qc.rs 251 MiB; PySCF 410 MiB. 39% less memory. Grid 23 × 33 × 39; 34.0 ų. lih-8 · 88 AO / cell · qc.rs 682 MiB; PySCF 1.02e+3 MiB. 33% less memory. Grid 51 × 45 × 45; 136.1 ų. lih-8 · 88 AO / cell · qc.rs 682 MiB; PySCF 1.02e+3 MiB. 33% less memory. Grid 51 × 45 × 45; 136.1 ų. lif-5 · 95 AO / cell · qc.rs 340 MiB; PySCF 759 MiB. 55% less memory. Grid 37 × 43 × 43; 81.2 ų. lif-5 · 95 AO / cell · qc.rs 340 MiB; PySCF 759 MiB. 55% less memory. Grid 37 × 43 × 43; 81.2 ų. lih-9 · 99 AO / cell · qc.rs 795 MiB; PySCF 1.21 GiB. 36% less memory. Grid 45 × 45 × 57; 153.2 ų. lih-9 · 99 AO / cell · qc.rs 795 MiB; PySCF 1.21 GiB. 36% less memory. Grid 45 × 45 × 57; 153.2 ų. c-4 · 104 AO / cell · qc.rs 318 MiB; PySCF 555 MiB. 43% less memory. Grid 33 × 33 × 33; 45.4 ų. c-4 · 104 AO / cell · qc.rs 318 MiB; PySCF 555 MiB. 43% less memory. Grid 33 × 33 × 33; 45.4 ų. lih-10 · 110 AO / cell · qc.rs 594 MiB; PySCF 1.46 GiB. 60% less memory. Grid 45 × 57 × 51; 170.2 ų. lih-10 · 110 AO / cell · qc.rs 594 MiB; PySCF 1.46 GiB. 60% less memory. Grid 45 × 57 × 51; 170.2 ų. lif-6 · 114 AO / cell · qc.rs 489 MiB; PySCF 1.00e+3 MiB. 51% less memory. Grid 43 × 43 × 43; 97.4 ų. lif-6 · 114 AO / cell · qc.rs 489 MiB; PySCF 1.00e+3 MiB. 51% less memory. Grid 43 × 43 × 43; 97.4 ų. c-5 · 130 AO / cell · qc.rs 470 MiB; PySCF 773 MiB. 39% less memory. Grid 39 × 33 × 39; 56.7 ų. c-5 · 130 AO / cell · qc.rs 470 MiB; PySCF 773 MiB. 39% less memory. Grid 39 × 33 × 39; 56.7 ų. lif-7 · 133 AO / cell · qc.rs 492 MiB; PySCF 1.11 GiB. 57% less memory. Grid 43 × 43 × 43; 113.7 ų. lif-7 · 133 AO / cell · qc.rs 492 MiB; PySCF 1.11 GiB. 57% less memory. Grid 43 × 43 × 43; 113.7 ų. lif-8 · 152 AO / cell · qc.rs 700 MiB; PySCF 1.46 GiB. 53% less memory. Grid 51 × 43 × 43; 129.9 ų. lif-8 · 152 AO / cell · qc.rs 700 MiB; PySCF 1.46 GiB. 53% less memory. Grid 51 × 43 × 43; 129.9 ų. c-6 · 156 AO / cell · qc.rs 608 MiB; PySCF 1.02 GiB. 42% less memory. Grid 39 × 39 × 39; 68.1 ų. c-6 · 156 AO / cell · qc.rs 608 MiB; PySCF 1.02 GiB. 42% less memory. Grid 39 × 39 × 39; 68.1 ų. lif-9 · 171 AO / cell · qc.rs 940 MiB; PySCF 1.80 GiB. 49% less memory. Grid 43 × 43 × 57; 146.2 ų. lif-9 · 171 AO / cell · qc.rs 940 MiB; PySCF 1.80 GiB. 49% less memory. Grid 43 × 43 × 57; 146.2 ų. c-7 · 182 AO / cell · qc.rs 684 MiB; PySCF 1.17 GiB. 43% less memory. Grid 39 × 39 × 39; 79.4 ų. c-7 · 182 AO / cell · qc.rs 684 MiB; PySCF 1.17 GiB. 43% less memory. Grid 39 × 39 × 39; 79.4 ų. lif-10 · 190 AO / cell · qc.rs 1.11 GiB; PySCF 2.29 GiB. 52% less memory. Grid 43 × 57 × 51; 162.4 ų. lif-10 · 190 AO / cell · qc.rs 1.11 GiB; PySCF 2.29 GiB. 52% less memory. Grid 43 × 57 × 51; 162.4 ų. c-8 · 208 AO / cell · qc.rs 855 MiB; PySCF 1.46 GiB. 43% less memory. Grid 45 × 39 × 39; 90.8 ų. c-8 · 208 AO / cell · qc.rs 855 MiB; PySCF 1.46 GiB. 43% less memory. Grid 45 × 39 × 39; 90.8 ų. c-9 · 234 AO / cell · qc.rs 927 MiB; PySCF 1.76 GiB. 49% less memory. Grid 39 × 39 × 49; 102.1 ų. c-9 · 234 AO / cell · qc.rs 927 MiB; PySCF 1.76 GiB. 49% less memory. Grid 39 × 39 × 49; 102.1 ų. c-10 · 260 AO / cell · qc.rs 1.11 GiB; PySCF 2.22 GiB. 50% less memory. Grid 39 × 49 × 45; 113.5 ų. c-10 · 260 AO / cell · qc.rs 1.11 GiB; PySCF 2.22 GiB. 50% less memory. Grid 39 × 49 × 45; 113.5 ų.
Caffeine 24 atoms · 246 AO · energy and forces · def2-SVP
PBE Hardware Time Peak RSS Apple M4 Max 9.1× faster
1.70 s qc.rs
15.5 s PySCF
90% less memory
394 MiB qc.rs
3.87 GiB PySCF
Apple M3 Ultra 8.3× faster
2.02 s qc.rs
16.8 s PySCF
72% less memory
1.04 GiB qc.rs
3.71 GiB PySCF
AMD EPYC 7702P 1.5× faster
7.14 s qc.rs
11.0 s PySCF
77% less memory
490 MiB qc.rs
2.07 GiB PySCF
r²SCAN Hardware Time Peak RSS Apple M4 Max 9.2× faster
2.00 s qc.rs
18.4 s PySCF
92% less memory
296 MiB qc.rs
3.54 GiB PySCF
Apple M3 Ultra 8.3× faster
2.35 s qc.rs
19.4 s PySCF
82% less memory
724 MiB qc.rs
3.83 GiB PySCF
AMD EPYC 7702P 1.9× faster
8.49 s qc.rs
16.3 s PySCF
77% less memory
490 MiB qc.rs
2.07 GiB PySCF
SKALA 1.1 Hardware Time Peak RSS Apple M4 Max 9.7× faster
8.49 s qc.rs
82.5 s PySCF + SKALA
81% less memory
3.98 GiB qc.rs
21.2 GiB PySCF + SKALA
Apple M3 Ultra 8.2× faster
12.1 s qc.rs
98.8 s PySCF + SKALA
75% less memory
5.11 GiB qc.rs
20.4 GiB PySCF + SKALA
AMD EPYC 7702P 2.9× faster
40.4 s qc.rs
119 s PySCF + SKALA
81% less memory
3.37 GiB qc.rs
18.1 GiB PySCF + SKALA
TiO₂ anatase 6 atoms · 104 AO / cell · energy, forces, and stress · Γ
PBE Hardware Time Peak RSS Apple M4 Max 10.2× faster
17.1 s qc.rs
174 s PySCF
86% less memory
1.38 GiB qc.rs
10.1 GiB PySCF
Apple M3 Ultra 9.5× faster
21.9 s qc.rs
209 s PySCF
77% less memory
2.33 GiB qc.rs
10.3 GiB PySCF
AMD EPYC 7702P 1.9× faster
53.1 s qc.rs
103 s PySCF
87% less memory
1.01 GiB qc.rs
7.83 GiB PySCF
r²SCAN Hardware Time Peak RSS Apple M4 Max 2.5× faster
17.2 s qc.rs
42.3 s CP2K
1.25 GiB qc.rs
686 MiB CP2K
Apple M3 Ultra 2.1× faster
22.0 s qc.rs
47.1 s CP2K
2.38 GiB qc.rs
1.14 GiB CP2K
AMD EPYC 7702P 1.2× faster
52.1 s qc.rs
64.4 s CP2K
989 MiB qc.rs
654 MiB CP2K
NVIDIA A100 · 40 GB · 8 host threads · Atomli 0.1.6 · GPU4PySCF 1.8.0
qc.rs GPU4PySCFLower is better
Molecules PBE / def2-SVP · RI-J · 30 AO sizes · logarithmic axes
Per SCF time 10.0 ms 100 ms 1.00 s 10 20 50 100 200 500 Atomic orbitals g2-H2 · 10 AO · qc.rs 28.2 ms; GPU4PySCF 89.3 ms. 3.2× faster. g2-H2 · 10 AO · qc.rs 28.2 ms; GPU4PySCF 89.3 ms. 3.2× faster. g2-CH2_s1A1d · 24 AO · qc.rs 11.1 ms; GPU4PySCF 66.4 ms. 6× faster. g2-CH2_s1A1d · 24 AO · qc.rs 11.1 ms; GPU4PySCF 66.4 ms. 6× faster. g2-NH3 · 29 AO · qc.rs 13.6 ms; GPU4PySCF 77.4 ms. 5.7× faster. g2-NH3 · 29 AO · qc.rs 13.6 ms; GPU4PySCF 77.4 ms. 5.7× faster. g2-CH4 · 34 AO · qc.rs 16.0 ms; GPU4PySCF 109 ms. 6.8× faster. g2-CH4 · 34 AO · qc.rs 16.0 ms; GPU4PySCF 109 ms. 6.8× faster. g2-HOCl · 37 AO · qc.rs 17.1 ms; GPU4PySCF 108 ms. 6.3× faster. g2-HOCl · 37 AO · qc.rs 17.1 ms; GPU4PySCF 108 ms. 6.3× faster. g2-C2H2 · 38 AO · qc.rs 16.3 ms; GPU4PySCF 108 ms. 6.6× faster. g2-C2H2 · 38 AO · qc.rs 16.3 ms; GPU4PySCF 108 ms. 6.6× faster. g2-ClNO · 46 AO · qc.rs 14.9 ms; GPU4PySCF 90.2 ms. 6.1× faster. g2-ClNO · 46 AO · qc.rs 14.9 ms; GPU4PySCF 90.2 ms. 6.1× faster. g2-C2H4 · 48 AO · qc.rs 21.4 ms; GPU4PySCF 122 ms. 5.7× faster. g2-C2H4 · 48 AO · qc.rs 21.4 ms; GPU4PySCF 122 ms. 5.7× faster. g2-H3CNH2 · 53 AO · qc.rs 17.8 ms; GPU4PySCF 95.2 ms. 5.3× faster. g2-H3CNH2 · 53 AO · qc.rs 17.8 ms; GPU4PySCF 95.2 ms. 5.3× faster. g2-CH3CN · 57 AO · qc.rs 17.8 ms; GPU4PySCF 88.3 ms. 5× faster. g2-CH3CN · 57 AO · qc.rs 17.8 ms; GPU4PySCF 88.3 ms. 5× faster. g2-C2H6 · 58 AO · qc.rs 19.8 ms; GPU4PySCF 126 ms. 6.4× faster. g2-C2H6 · 58 AO · qc.rs 19.8 ms; GPU4PySCF 126 ms. 6.4× faster. g2-H2CCHCl · 61 AO · qc.rs 17.5 ms; GPU4PySCF 99.3 ms. 5.7× faster. g2-H2CCHCl · 61 AO · qc.rs 17.5 ms; GPU4PySCF 99.3 ms. 5.7× faster. g2-CH2SCH2 · 66 AO · qc.rs 25.1 ms; GPU4PySCF 116 ms. 4.6× faster. g2-CH2SCH2 · 66 AO · qc.rs 25.1 ms; GPU4PySCF 116 ms. 4.6× faster. g2-CH3CH2Cl · 71 AO · qc.rs 24.5 ms; GPU4PySCF 123 ms. 5× faster. g2-CH3CH2Cl · 71 AO · qc.rs 24.5 ms; GPU4PySCF 123 ms. 5× faster. g2-HCCl3 · 73 AO · qc.rs 23.6 ms; GPU4PySCF 116 ms. 4.9× faster. g2-HCCl3 · 73 AO · qc.rs 23.6 ms; GPU4PySCF 116 ms. 4.9× faster. g2-CH3CH2SH · 76 AO · qc.rs 25.9 ms; GPU4PySCF 125 ms. 4.8× faster. g2-CH3CH2SH · 76 AO · qc.rs 25.9 ms; GPU4PySCF 125 ms. 4.8× faster. g2-C2H6NH · 77 AO · qc.rs 24.4 ms; GPU4PySCF 96.2 ms. 3.9× faster. g2-C2H6NH · 77 AO · qc.rs 24.4 ms; GPU4PySCF 96.2 ms. 3.9× faster. g2-CH3CONH2 · 81 AO · qc.rs 22.6 ms; GPU4PySCF 96.2 ms. 4.3× faster. g2-CH3CONH2 · 81 AO · qc.rs 22.6 ms; GPU4PySCF 96.2 ms. 4.3× faster. g2-2-butyne · 86 AO · qc.rs 24.3 ms; GPU4PySCF 129 ms. 5.3× faster. g2-2-butyne · 86 AO · qc.rs 24.3 ms; GPU4PySCF 129 ms. 5.3× faster. g2-C4H4S · 94 AO · qc.rs 25.0 ms; GPU4PySCF 110 ms. 4.4× faster. g2-C4H4S · 94 AO · qc.rs 25.0 ms; GPU4PySCF 110 ms. 4.4× faster. g2-C2Cl4 · 100 AO · qc.rs 26.2 ms; GPU4PySCF 107 ms. 4.1× faster. g2-C2Cl4 · 100 AO · qc.rs 26.2 ms; GPU4PySCF 107 ms. 4.1× faster. g2-isobutane · 106 AO · qc.rs 33.7 ms; GPU4PySCF 121 ms. 3.6× faster. g2-isobutane · 106 AO · qc.rs 33.7 ms; GPU4PySCF 121 ms. 3.6× faster. g2-C6H6 · 114 AO · qc.rs 29.1 ms; GPU4PySCF 105 ms. 3.6× faster. g2-C6H6 · 114 AO · qc.rs 29.1 ms; GPU4PySCF 105 ms. 3.6× faster. alkane-6 · 154 AO · qc.rs 48.7 ms; GPU4PySCF 127 ms. 2.6× faster. alkane-6 · 154 AO · qc.rs 48.7 ms; GPU4PySCF 127 ms. 2.6× faster. alkane-8 · 202 AO · qc.rs 74.4 ms; GPU4PySCF 139 ms. 1.9× faster. alkane-8 · 202 AO · qc.rs 74.4 ms; GPU4PySCF 139 ms. 1.9× faster. caffeine · 246 AO · qc.rs 85.5 ms; GPU4PySCF 147 ms. 1.7× faster. caffeine · 246 AO · qc.rs 85.5 ms; GPU4PySCF 147 ms. 1.7× faster. alkane-10 · 250 AO · qc.rs 104 ms; GPU4PySCF 155 ms. 1.5× faster. alkane-10 · 250 AO · qc.rs 104 ms; GPU4PySCF 155 ms. 1.5× faster. alkane-12 · 298 AO · qc.rs 157 ms; GPU4PySCF 173 ms. 1.1× faster. alkane-12 · 298 AO · qc.rs 157 ms; GPU4PySCF 173 ms. 1.1× faster. alkane-16 · 394 AO · qc.rs 350 ms; GPU4PySCF 210 ms. 1.7× slower. alkane-16 · 394 AO · qc.rs 350 ms; GPU4PySCF 210 ms. 1.7× slower. alkane-20 · 490 AO · qc.rs 535 ms; GPU4PySCF 245 ms. 2.2× slower. alkane-20 · 490 AO · qc.rs 535 ms; GPU4PySCF 245 ms. 2.2× slower. Peak host RSS 95.4 MiB 954 MiB 10 20 50 100 200 500 Atomic orbitals g2-H2 · 10 AO · qc.rs 368 MiB; GPU4PySCF 911 MiB. 60% less memory. g2-H2 · 10 AO · qc.rs 368 MiB; GPU4PySCF 911 MiB. 60% less memory. g2-CH2_s1A1d · 24 AO · qc.rs 386 MiB; GPU4PySCF 873 MiB. 56% less memory. g2-CH2_s1A1d · 24 AO · qc.rs 386 MiB; GPU4PySCF 873 MiB. 56% less memory. g2-NH3 · 29 AO · qc.rs 386 MiB; GPU4PySCF 873 MiB. 56% less memory. g2-NH3 · 29 AO · qc.rs 386 MiB; GPU4PySCF 873 MiB. 56% less memory. g2-CH4 · 34 AO · qc.rs 393 MiB; GPU4PySCF 873 MiB. 55% less memory. g2-CH4 · 34 AO · qc.rs 393 MiB; GPU4PySCF 873 MiB. 55% less memory. g2-HOCl · 37 AO · qc.rs 398 MiB; GPU4PySCF 874 MiB. 54% less memory. g2-HOCl · 37 AO · qc.rs 398 MiB; GPU4PySCF 874 MiB. 54% less memory. g2-C2H2 · 38 AO · qc.rs 398 MiB; GPU4PySCF 873 MiB. 54% less memory. g2-C2H2 · 38 AO · qc.rs 398 MiB; GPU4PySCF 873 MiB. 54% less memory. g2-ClNO · 46 AO · qc.rs 404 MiB; GPU4PySCF 868 MiB. 53% less memory. g2-ClNO · 46 AO · qc.rs 404 MiB; GPU4PySCF 868 MiB. 53% less memory. g2-C2H4 · 48 AO · qc.rs 395 MiB; GPU4PySCF 875 MiB. 55% less memory. g2-C2H4 · 48 AO · qc.rs 395 MiB; GPU4PySCF 875 MiB. 55% less memory. g2-H3CNH2 · 53 AO · qc.rs 399 MiB; GPU4PySCF 874 MiB. 54% less memory. g2-H3CNH2 · 53 AO · qc.rs 399 MiB; GPU4PySCF 874 MiB. 54% less memory. g2-CH3CN · 57 AO · qc.rs 399 MiB; GPU4PySCF 882 MiB. 55% less memory. g2-CH3CN · 57 AO · qc.rs 399 MiB; GPU4PySCF 882 MiB. 55% less memory. g2-C2H6 · 58 AO · qc.rs 402 MiB; GPU4PySCF 875 MiB. 54% less memory. g2-C2H6 · 58 AO · qc.rs 402 MiB; GPU4PySCF 875 MiB. 54% less memory. g2-H2CCHCl · 61 AO · qc.rs 404 MiB; GPU4PySCF 883 MiB. 54% less memory. g2-H2CCHCl · 61 AO · qc.rs 404 MiB; GPU4PySCF 883 MiB. 54% less memory. g2-CH2SCH2 · 66 AO · qc.rs 414 MiB; GPU4PySCF 875 MiB. 53% less memory. g2-CH2SCH2 · 66 AO · qc.rs 414 MiB; GPU4PySCF 875 MiB. 53% less memory. g2-CH3CH2Cl · 71 AO · qc.rs 410 MiB; GPU4PySCF 882 MiB. 53% less memory. g2-CH3CH2Cl · 71 AO · qc.rs 410 MiB; GPU4PySCF 882 MiB. 53% less memory. g2-HCCl3 · 73 AO · qc.rs 415 MiB; GPU4PySCF 875 MiB. 53% less memory. g2-HCCl3 · 73 AO · qc.rs 415 MiB; GPU4PySCF 875 MiB. 53% less memory. g2-CH3CH2SH · 76 AO · qc.rs 410 MiB; GPU4PySCF 875 MiB. 53% less memory. g2-CH3CH2SH · 76 AO · qc.rs 410 MiB; GPU4PySCF 875 MiB. 53% less memory. g2-C2H6NH · 77 AO · qc.rs 409 MiB; GPU4PySCF 881 MiB. 54% less memory. g2-C2H6NH · 77 AO · qc.rs 409 MiB; GPU4PySCF 881 MiB. 54% less memory. g2-CH3CONH2 · 81 AO · qc.rs 414 MiB; GPU4PySCF 877 MiB. 53% less memory. g2-CH3CONH2 · 81 AO · qc.rs 414 MiB; GPU4PySCF 877 MiB. 53% less memory. g2-2-butyne · 86 AO · qc.rs 405 MiB; GPU4PySCF 874 MiB. 54% less memory. g2-2-butyne · 86 AO · qc.rs 405 MiB; GPU4PySCF 874 MiB. 54% less memory. g2-C4H4S · 94 AO · qc.rs 421 MiB; GPU4PySCF 883 MiB. 52% less memory. g2-C4H4S · 94 AO · qc.rs 421 MiB; GPU4PySCF 883 MiB. 52% less memory. g2-C2Cl4 · 100 AO · qc.rs 425 MiB; GPU4PySCF 883 MiB. 52% less memory. g2-C2Cl4 · 100 AO · qc.rs 425 MiB; GPU4PySCF 883 MiB. 52% less memory. g2-isobutane · 106 AO · qc.rs 426 MiB; GPU4PySCF 874 MiB. 51% less memory. g2-isobutane · 106 AO · qc.rs 426 MiB; GPU4PySCF 874 MiB. 51% less memory. g2-C6H6 · 114 AO · qc.rs 428 MiB; GPU4PySCF 887 MiB. 52% less memory. g2-C6H6 · 114 AO · qc.rs 428 MiB; GPU4PySCF 887 MiB. 52% less memory. alkane-6 · 154 AO · qc.rs 433 MiB; GPU4PySCF 891 MiB. 51% less memory. alkane-6 · 154 AO · qc.rs 433 MiB; GPU4PySCF 891 MiB. 51% less memory. alkane-8 · 202 AO · qc.rs 469 MiB; GPU4PySCF 892 MiB. 47% less memory. alkane-8 · 202 AO · qc.rs 469 MiB; GPU4PySCF 892 MiB. 47% less memory. caffeine · 246 AO · qc.rs 514 MiB; GPU4PySCF 886 MiB. 42% less memory. caffeine · 246 AO · qc.rs 514 MiB; GPU4PySCF 886 MiB. 42% less memory. alkane-10 · 250 AO · qc.rs 482 MiB; GPU4PySCF 893 MiB. 46% less memory. alkane-10 · 250 AO · qc.rs 482 MiB; GPU4PySCF 893 MiB. 46% less memory. alkane-12 · 298 AO · qc.rs 521 MiB; GPU4PySCF 894 MiB. 42% less memory. alkane-12 · 298 AO · qc.rs 521 MiB; GPU4PySCF 894 MiB. 42% less memory. alkane-16 · 394 AO · qc.rs 588 MiB; GPU4PySCF 920 MiB. 36% less memory. alkane-16 · 394 AO · qc.rs 588 MiB; GPU4PySCF 920 MiB. 36% less memory. alkane-20 · 490 AO · qc.rs 591 MiB; GPU4PySCF 896 MiB. 34% less memory. alkane-20 · 490 AO · qc.rs 591 MiB; GPU4PySCF 896 MiB. 34% less memory.
Periodic cells PBE / GTH-DZVP-MOLOPT-SR · Γ sampling · 30 AO sizes · logarithmic axes
Per SCF time 10.0 ms 100 ms 1.00 s 10.0 s 10 20 50 100 200 AO per cell lih-1 · 11 AO / cell · qc.rs 255 ms; GPU4PySCF 6.93 s. 27.2× faster. Grid 27 × 27 × 27; 17.0 ų. lih-1 · 11 AO / cell · qc.rs 255 ms; GPU4PySCF 6.93 s. 27.2× faster. Grid 27 × 27 × 27; 17.0 ų. lif-1 · 19 AO / cell · qc.rs 141 ms; GPU4PySCF 9.93 s. 70.6× faster. Grid 27 × 27 × 27; 16.2 ų. lif-1 · 19 AO / cell · qc.rs 141 ms; GPU4PySCF 9.93 s. 70.6× faster. Grid 27 × 27 × 27; 16.2 ų. lih-2 · 22 AO / cell · qc.rs 125 ms; GPU4PySCF 2.60 s. 20.8× faster. Grid 27 × 37 × 27; 34.0 ų. lih-2 · 22 AO / cell · qc.rs 125 ms; GPU4PySCF 2.60 s. 20.8× faster. Grid 27 × 37 × 27; 34.0 ų. oli2-1 · 25 AO / cell · qc.rs 150 ms; GPU4PySCF 9.80 s. 65.4× faster. Grid 29 × 29 × 29; 24.7 ų. oli2-1 · 25 AO / cell · qc.rs 150 ms; GPU4PySCF 9.80 s. 65.4× faster. Grid 29 × 29 × 29; 24.7 ų. c-1 · 26 AO / cell · qc.rs 60.8 ms; GPU4PySCF 1.12 s. 18.4× faster. Grid 23 × 23 × 23; 11.3 ų. c-1 · 26 AO / cell · qc.rs 60.8 ms; GPU4PySCF 1.12 s. 18.4× faster. Grid 23 × 23 × 23; 11.3 ų. kf-1 · 27 AO / cell · qc.rs 184 ms; GPU4PySCF 6.68 s. 36.4× faster. Grid 35 × 35 × 35; 38.5 ų. kf-1 · 27 AO / cell · qc.rs 184 ms; GPU4PySCF 6.68 s. 36.4× faster. Grid 35 × 35 × 35; 38.5 ų. lih-3 · 33 AO / cell · qc.rs 175 ms; GPU4PySCF 2.71 s. 15.4× faster. Grid 27 × 37 × 45; 51.1 ų. lih-3 · 33 AO / cell · qc.rs 175 ms; GPU4PySCF 2.71 s. 15.4× faster. Grid 27 × 37 × 45; 51.1 ų. lif-2 · 38 AO / cell · qc.rs 116 ms; GPU4PySCF 4.17 s. 35.8× faster. Grid 27 × 37 × 27; 32.5 ų. lif-2 · 38 AO / cell · qc.rs 116 ms; GPU4PySCF 4.17 s. 35.8× faster. Grid 27 × 37 × 27; 32.5 ų. lih-4 · 44 AO / cell · qc.rs 174 ms; GPU4PySCF 1.28 s. 7.3× faster. Grid 37 × 37 × 37; 68.1 ų. lih-4 · 44 AO / cell · qc.rs 174 ms; GPU4PySCF 1.28 s. 7.3× faster. Grid 37 × 37 × 37; 68.1 ų. c-2 · 52 AO / cell · qc.rs 68.5 ms; GPU4PySCF 485 ms. 7.1× faster. Grid 23 × 33 × 23; 22.7 ų. c-2 · 52 AO / cell · qc.rs 68.5 ms; GPU4PySCF 485 ms. 7.1× faster. Grid 23 × 33 × 23; 22.7 ų. lih-5 · 55 AO / cell · qc.rs 249 ms; GPU4PySCF 3.12 s. 12.5× faster. Grid 37 × 45 × 45; 85.1 ų. lih-5 · 55 AO / cell · qc.rs 249 ms; GPU4PySCF 3.12 s. 12.5× faster. Grid 37 × 45 × 45; 85.1 ų. lif-3 · 57 AO / cell · qc.rs 211 ms; GPU4PySCF 1.38 s. 6.5× faster. Grid 27 × 37 × 43; 48.7 ų. lif-3 · 57 AO / cell · qc.rs 211 ms; GPU4PySCF 1.38 s. 6.5× faster. Grid 27 × 37 × 43; 48.7 ų. lih-6 · 66 AO / cell · qc.rs 298 ms; GPU4PySCF 3.46 s. 11.6× faster. Grid 45 × 45 × 45; 102.1 ų. lih-6 · 66 AO / cell · qc.rs 298 ms; GPU4PySCF 3.46 s. 11.6× faster. Grid 45 × 45 × 45; 102.1 ų. lif-4 · 76 AO / cell · qc.rs 190 ms; GPU4PySCF 2.24 s. 11.8× faster. Grid 37 × 37 × 37; 65.0 ų. lif-4 · 76 AO / cell · qc.rs 190 ms; GPU4PySCF 2.24 s. 11.8× faster. Grid 37 × 37 × 37; 65.0 ų. lih-7 · 77 AO / cell · qc.rs 318 ms; GPU4PySCF 3.43 s. 10.8× faster. Grid 45 × 45 × 45; 119.1 ų. lih-7 · 77 AO / cell · qc.rs 318 ms; GPU4PySCF 3.43 s. 10.8× faster. Grid 45 × 45 × 45; 119.1 ų. c-3 · 78 AO / cell · qc.rs 136 ms; GPU4PySCF 415 ms. 3× faster. Grid 23 × 33 × 39; 34.0 ų. c-3 · 78 AO / cell · qc.rs 136 ms; GPU4PySCF 415 ms. 3× faster. Grid 23 × 33 × 39; 34.0 ų. lih-8 · 88 AO / cell · qc.rs 366 ms; GPU4PySCF 2.93 s. 8× faster. Grid 51 × 45 × 45; 136.1 ų. lih-8 · 88 AO / cell · qc.rs 366 ms; GPU4PySCF 2.93 s. 8× faster. Grid 51 × 45 × 45; 136.1 ų. lif-5 · 95 AO / cell · qc.rs 293 ms; GPU4PySCF 4.07 s. 13.9× faster. Grid 37 × 43 × 43; 81.2 ų. lif-5 · 95 AO / cell · qc.rs 293 ms; GPU4PySCF 4.07 s. 13.9× faster. Grid 37 × 43 × 43; 81.2 ų. lih-9 · 99 AO / cell · qc.rs 435 ms; GPU4PySCF 2.97 s. 6.8× faster. Grid 45 × 45 × 57; 153.2 ų. lih-9 · 99 AO / cell · qc.rs 435 ms; GPU4PySCF 2.97 s. 6.8× faster. Grid 45 × 45 × 57; 153.2 ų. c-4 · 104 AO / cell · qc.rs 211 ms; GPU4PySCF 268 ms. 1.3× faster. Grid 33 × 33 × 33; 45.4 ų. c-4 · 104 AO / cell · qc.rs 211 ms; GPU4PySCF 268 ms. 1.3× faster. Grid 33 × 33 × 33; 45.4 ų. lih-10 · 110 AO / cell · qc.rs 515 ms; GPU4PySCF 3.20 s. 6.2× faster. Grid 45 × 57 × 51; 170.2 ų. lih-10 · 110 AO / cell · qc.rs 515 ms; GPU4PySCF 3.20 s. 6.2× faster. Grid 45 × 57 × 51; 170.2 ų. lif-6 · 114 AO / cell · qc.rs 475 ms; GPU4PySCF 4.45 s. 9.4× faster. Grid 43 × 43 × 43; 97.4 ų. lif-6 · 114 AO / cell · qc.rs 475 ms; GPU4PySCF 4.45 s. 9.4× faster. Grid 43 × 43 × 43; 97.4 ų. c-5 · 130 AO / cell · qc.rs 379 ms; GPU4PySCF 483 ms. 1.3× faster. Grid 39 × 33 × 39; 56.7 ų. c-5 · 130 AO / cell · qc.rs 379 ms; GPU4PySCF 483 ms. 1.3× faster. Grid 39 × 33 × 39; 56.7 ų. lif-7 · 133 AO / cell · qc.rs 516 ms; GPU4PySCF 4.42 s. 8.6× faster. Grid 43 × 43 × 43; 113.7 ų. lif-7 · 133 AO / cell · qc.rs 516 ms; GPU4PySCF 4.42 s. 8.6× faster. Grid 43 × 43 × 43; 113.7 ų. lif-8 · 152 AO / cell · qc.rs 736 ms; GPU4PySCF 1.45 s. 2× faster. Grid 51 × 43 × 43; 129.9 ų. lif-8 · 152 AO / cell · qc.rs 736 ms; GPU4PySCF 1.45 s. 2× faster. Grid 51 × 43 × 43; 129.9 ų. c-6 · 156 AO / cell · qc.rs 600 ms; GPU4PySCF 558 ms. 1.08× slower. Grid 39 × 39 × 39; 68.1 ų. c-6 · 156 AO / cell · qc.rs 600 ms; GPU4PySCF 558 ms. 1.08× slower. Grid 39 × 39 × 39; 68.1 ų. lif-9 · 171 AO / cell · qc.rs 1.07 s; GPU4PySCF 1.17 s. 1.09× faster. Grid 43 × 43 × 57; 146.2 ų. lif-9 · 171 AO / cell · qc.rs 1.07 s; GPU4PySCF 1.17 s. 1.09× faster. Grid 43 × 43 × 57; 146.2 ų. c-7 · 182 AO / cell · qc.rs 1.13 s; GPU4PySCF 534 ms. 2.1× slower. Grid 39 × 39 × 39; 79.4 ų. c-7 · 182 AO / cell · qc.rs 1.13 s; GPU4PySCF 534 ms. 2.1× slower. Grid 39 × 39 × 39; 79.4 ų. lif-10 · 190 AO / cell · qc.rs 1.71 s; GPU4PySCF 1.61 s. 1.06× slower. Grid 43 × 57 × 51; 162.4 ų. lif-10 · 190 AO / cell · qc.rs 1.71 s; GPU4PySCF 1.61 s. 1.06× slower. Grid 43 × 57 × 51; 162.4 ų. c-8 · 208 AO / cell · qc.rs 1.65 s; GPU4PySCF 512 ms. 3.2× slower. Grid 45 × 39 × 39; 90.8 ų. c-8 · 208 AO / cell · qc.rs 1.65 s; GPU4PySCF 512 ms. 3.2× slower. Grid 45 × 39 × 39; 90.8 ų. Peak host RSS 95.4 MiB 954 MiB 9.31 GiB 10 20 50 100 200 AO per cell lih-1 · 11 AO / cell · qc.rs 348 MiB; GPU4PySCF 1.30 GiB. 74% less memory. Grid 27 × 27 × 27; 17.0 ų. lih-1 · 11 AO / cell · qc.rs 348 MiB; GPU4PySCF 1.30 GiB. 74% less memory. Grid 27 × 27 × 27; 17.0 ų. lif-1 · 19 AO / cell · qc.rs 350 MiB; GPU4PySCF 1.28 GiB. 73% less memory. Grid 27 × 27 × 27; 16.2 ų. lif-1 · 19 AO / cell · qc.rs 350 MiB; GPU4PySCF 1.28 GiB. 73% less memory. Grid 27 × 27 × 27; 16.2 ų. lih-2 · 22 AO / cell · qc.rs 355 MiB; GPU4PySCF 1.28 GiB. 73% less memory. Grid 27 × 37 × 27; 34.0 ų. lih-2 · 22 AO / cell · qc.rs 355 MiB; GPU4PySCF 1.28 GiB. 73% less memory. Grid 27 × 37 × 27; 34.0 ų. oli2-1 · 25 AO / cell · qc.rs 354 MiB; GPU4PySCF 1.26 GiB. 73% less memory. Grid 29 × 29 × 29; 24.7 ų. oli2-1 · 25 AO / cell · qc.rs 354 MiB; GPU4PySCF 1.26 GiB. 73% less memory. Grid 29 × 29 × 29; 24.7 ų. c-1 · 26 AO / cell · qc.rs 348 MiB; GPU4PySCF 1.25 GiB. 73% less memory. Grid 23 × 23 × 23; 11.3 ų. c-1 · 26 AO / cell · qc.rs 348 MiB; GPU4PySCF 1.25 GiB. 73% less memory. Grid 23 × 23 × 23; 11.3 ų. kf-1 · 27 AO / cell · qc.rs 356 MiB; GPU4PySCF 1.26 GiB. 73% less memory. Grid 35 × 35 × 35; 38.5 ų. kf-1 · 27 AO / cell · qc.rs 356 MiB; GPU4PySCF 1.26 GiB. 73% less memory. Grid 35 × 35 × 35; 38.5 ų. lih-3 · 33 AO / cell · qc.rs 355 MiB; GPU4PySCF 1.28 GiB. 73% less memory. Grid 27 × 37 × 45; 51.1 ų. lih-3 · 33 AO / cell · qc.rs 355 MiB; GPU4PySCF 1.28 GiB. 73% less memory. Grid 27 × 37 × 45; 51.1 ų. lif-2 · 38 AO / cell · qc.rs 353 MiB; GPU4PySCF 1.26 GiB. 73% less memory. Grid 27 × 37 × 27; 32.5 ų. lif-2 · 38 AO / cell · qc.rs 353 MiB; GPU4PySCF 1.26 GiB. 73% less memory. Grid 27 × 37 × 27; 32.5 ų. lih-4 · 44 AO / cell · qc.rs 366 MiB; GPU4PySCF 1.29 GiB. 72% less memory. Grid 37 × 37 × 37; 68.1 ų. lih-4 · 44 AO / cell · qc.rs 366 MiB; GPU4PySCF 1.29 GiB. 72% less memory. Grid 37 × 37 × 37; 68.1 ų. c-2 · 52 AO / cell · qc.rs 350 MiB; GPU4PySCF 1.26 GiB. 73% less memory. Grid 23 × 33 × 23; 22.7 ų. c-2 · 52 AO / cell · qc.rs 350 MiB; GPU4PySCF 1.26 GiB. 73% less memory. Grid 23 × 33 × 23; 22.7 ų. lih-5 · 55 AO / cell · qc.rs 365 MiB; GPU4PySCF 1.29 GiB. 72% less memory. Grid 37 × 45 × 45; 85.1 ų. lih-5 · 55 AO / cell · qc.rs 365 MiB; GPU4PySCF 1.29 GiB. 72% less memory. Grid 37 × 45 × 45; 85.1 ų. lif-3 · 57 AO / cell · qc.rs 357 MiB; GPU4PySCF 1.28 GiB. 73% less memory. Grid 27 × 37 × 43; 48.7 ų. lif-3 · 57 AO / cell · qc.rs 357 MiB; GPU4PySCF 1.28 GiB. 73% less memory. Grid 27 × 37 × 43; 48.7 ų. lih-6 · 66 AO / cell · qc.rs 369 MiB; GPU4PySCF 1.28 GiB. 72% less memory. Grid 45 × 45 × 45; 102.1 ų. lih-6 · 66 AO / cell · qc.rs 369 MiB; GPU4PySCF 1.28 GiB. 72% less memory. Grid 45 × 45 × 45; 102.1 ų. lif-4 · 76 AO / cell · qc.rs 362 MiB; GPU4PySCF 1.29 GiB. 73% less memory. Grid 37 × 37 × 37; 65.0 ų. lif-4 · 76 AO / cell · qc.rs 362 MiB; GPU4PySCF 1.29 GiB. 73% less memory. Grid 37 × 37 × 37; 65.0 ų. lih-7 · 77 AO / cell · qc.rs 369 MiB; GPU4PySCF 1.28 GiB. 72% less memory. Grid 45 × 45 × 45; 119.1 ų. lih-7 · 77 AO / cell · qc.rs 369 MiB; GPU4PySCF 1.28 GiB. 72% less memory. Grid 45 × 45 × 45; 119.1 ų. c-3 · 78 AO / cell · qc.rs 354 MiB; GPU4PySCF 1.28 GiB. 73% less memory. Grid 23 × 33 × 39; 34.0 ų. c-3 · 78 AO / cell · qc.rs 354 MiB; GPU4PySCF 1.28 GiB. 73% less memory. Grid 23 × 33 × 39; 34.0 ų. lih-8 · 88 AO / cell · qc.rs 372 MiB; GPU4PySCF 1.28 GiB. 72% less memory. Grid 51 × 45 × 45; 136.1 ų. lih-8 · 88 AO / cell · qc.rs 372 MiB; GPU4PySCF 1.28 GiB. 72% less memory. Grid 51 × 45 × 45; 136.1 ų. lif-5 · 95 AO / cell · qc.rs 371 MiB; GPU4PySCF 1.28 GiB. 72% less memory. Grid 37 × 43 × 43; 81.2 ų. lif-5 · 95 AO / cell · qc.rs 371 MiB; GPU4PySCF 1.28 GiB. 72% less memory. Grid 37 × 43 × 43; 81.2 ų. lih-9 · 99 AO / cell · qc.rs 375 MiB; GPU4PySCF 1.30 GiB. 72% less memory. Grid 45 × 45 × 57; 153.2 ų. lih-9 · 99 AO / cell · qc.rs 375 MiB; GPU4PySCF 1.30 GiB. 72% less memory. Grid 45 × 45 × 57; 153.2 ų. c-4 · 104 AO / cell · qc.rs 358 MiB; GPU4PySCF 1.28 GiB. 73% less memory. Grid 33 × 33 × 33; 45.4 ų. c-4 · 104 AO / cell · qc.rs 358 MiB; GPU4PySCF 1.28 GiB. 73% less memory. Grid 33 × 33 × 33; 45.4 ų. lih-10 · 110 AO / cell · qc.rs 378 MiB; GPU4PySCF 1.30 GiB. 72% less memory. Grid 45 × 57 × 51; 170.2 ų. lih-10 · 110 AO / cell · qc.rs 378 MiB; GPU4PySCF 1.30 GiB. 72% less memory. Grid 45 × 57 × 51; 170.2 ų. lif-6 · 114 AO / cell · qc.rs 378 MiB; GPU4PySCF 1.27 GiB. 71% less memory. Grid 43 × 43 × 43; 97.4 ų. lif-6 · 114 AO / cell · qc.rs 378 MiB; GPU4PySCF 1.27 GiB. 71% less memory. Grid 43 × 43 × 43; 97.4 ų. c-5 · 130 AO / cell · qc.rs 364 MiB; GPU4PySCF 1.30 GiB. 73% less memory. Grid 39 × 33 × 39; 56.7 ų. c-5 · 130 AO / cell · qc.rs 364 MiB; GPU4PySCF 1.30 GiB. 73% less memory. Grid 39 × 33 × 39; 56.7 ų. lif-7 · 133 AO / cell · qc.rs 385 MiB; GPU4PySCF 1.27 GiB. 70% less memory. Grid 43 × 43 × 43; 113.7 ų. lif-7 · 133 AO / cell · qc.rs 385 MiB; GPU4PySCF 1.27 GiB. 70% less memory. Grid 43 × 43 × 43; 113.7 ų. lif-8 · 152 AO / cell · qc.rs 398 MiB; GPU4PySCF 1.28 GiB. 70% less memory. Grid 51 × 43 × 43; 129.9 ų. lif-8 · 152 AO / cell · qc.rs 398 MiB; GPU4PySCF 1.28 GiB. 70% less memory. Grid 51 × 43 × 43; 129.9 ų. c-6 · 156 AO / cell · qc.rs 371 MiB; GPU4PySCF 1.27 GiB. 71% less memory. Grid 39 × 39 × 39; 68.1 ų. c-6 · 156 AO / cell · qc.rs 371 MiB; GPU4PySCF 1.27 GiB. 71% less memory. Grid 39 × 39 × 39; 68.1 ų. lif-9 · 171 AO / cell · qc.rs 411 MiB; GPU4PySCF 1.28 GiB. 69% less memory. Grid 43 × 43 × 57; 146.2 ų. lif-9 · 171 AO / cell · qc.rs 411 MiB; GPU4PySCF 1.28 GiB. 69% less memory. Grid 43 × 43 × 57; 146.2 ų. c-7 · 182 AO / cell · qc.rs 376 MiB; GPU4PySCF 1.26 GiB. 71% less memory. Grid 39 × 39 × 39; 79.4 ų. c-7 · 182 AO / cell · qc.rs 376 MiB; GPU4PySCF 1.26 GiB. 71% less memory. Grid 39 × 39 × 39; 79.4 ų. lif-10 · 190 AO / cell · qc.rs 452 MiB; GPU4PySCF 1.28 GiB. 66% less memory. Grid 43 × 57 × 51; 162.4 ų. lif-10 · 190 AO / cell · qc.rs 452 MiB; GPU4PySCF 1.28 GiB. 66% less memory. Grid 43 × 57 × 51; 162.4 ų. c-8 · 208 AO / cell · qc.rs 388 MiB; GPU4PySCF 1.26 GiB. 70% less memory. Grid 45 × 39 × 39; 90.8 ų. c-8 · 208 AO / cell · qc.rs 388 MiB; GPU4PySCF 1.26 GiB. 70% less memory. Grid 45 × 39 × 39; 90.8 ų.
Median of three changed-geometry evaluations after one warm-up. Each energy-only time is divided by its SCF iteration count.
Download graph (SVG) · Download measurements (JSON)
NVIDIA A100 · 40 GB Atomli 0.1.5 · development build · 8 host threads
Caffeine 24 atoms · 246 AO · energy + forces
Method Time Peak RSS PBE 1.3× faster
717 ms qc.rs
927 ms GPU4PySCF
46% less memory
657 MiB qc.rs
1.19 GiB GPU4PySCF
r²SCAN 1.7× faster
710 ms qc.rs
1.22 s GPU4PySCF
36% less memory
782 MiB qc.rs
1.19 GiB GPU4PySCF
SKALA 1.1 1.7× faster
1.56 s qc.rs
2.69 s SKALA / GPU4PySCF
71% less memory
666 MiB qc.rs
2.22 GiB SKALA / GPU4PySCF
TiO₂ anatase 6 atoms · 104 AO / cell · energy + forces + stress
Method Time Peak RSS PBE 1.6× faster
7.02 s qc.rs
11.2 s GPU4PySCF
64% less memory
551 MiB qc.rs
1.48 GiB GPU4PySCF
Methods GPU4PySCF 1.8.0 for PBE and r²SCAN. SKALA 2026.8 uses GPU4PySCF 1.6.1. Both use cuTENSOR 2.3.1.
Times are medians of three evaluations after one initial calculation. Geometry changes retain the preceding density.
Apple M4 Max · Metal Atomli 0.1.4 · PBE · 8 host threads
Energy + forces · TiO₂ includes stress
Methods and fixtures Per SCF time is the median of three ratios: each calculation time, including geometry setup, divided by its measured iteration count.
Complete calculation time is the median of three evaluations, including forces and periodic stress.
The measurements follow one converged warm-up.
Molecules use PBE/def2-SVP: ASE G2 geometries, caffeine, and all-trans alkanes.
Periodic AO counts are per cell. Grid size and k-point sampling also affect periodic cost.
The periodic sweep uses Γ sampling and a 100 Hartree grid resolution setting.
Periodic fixture definitions · GTH-DZVP-MOLOPT-SR · Γ sampling Material Atoms / cell AO / cell Grid points LiH 2–20 11–110 19,683–130,815 LiF 2–20 19–190 19,683–125,001 Diamond 2–20 26–260 12,167–85,995 KF 2–2 27–27 42,875–42,875 Li₂O 3–3 25–25 24,389–24,389
Warm calls: three coordinate changes of 0.001, 0.002, and 0.003 angstrom after one converged warm-up. Both engines retain the preceding density.
SCF limit: 50 iterations for qc.rs and PySCF. CP2K uses 50 inner optimizer steps and two outer iterations.
Molecular thresholds: qc.rs energy 10⁻⁹ Hartree, density 10⁻⁶; PySCF conv_tol=1e-6.
Periodic thresholds: qc.rs precision=1e-6; PySCF conv_tol=1e-6; CP2K EPS_SCF=1e-6.
TiO₂: qc.rs and PySCF use 104 AO and a 49³ grid. CP2K uses GPW, an 800 Ry cutoff, and an 80 Ry relative cutoff.